The 309 Doesn't Always Leave Elmhurst

I catch the Pace 309 at the Elmhurst Metra station. It's the start of the line: every eastbound 309 trip begins there, at 1st Street and York, and heads out along Lake Street to Austin Boulevard. More than once this summer I stood at that stop on a weekday afternoon for over an hour with no bus. At that time of day, Pace's schedule says one should leave every half hour.

I wanted to know whether that was bad luck or a pattern. Pace doesn't publish any record of where its buses actually went, so I started keeping one.

How I got the data

Pace's Bus Tracker website shows live bus positions on a map. Behind that map is a request the page makes to Pace's server, route by route, that returns every bus currently on the route with its location. It needs no login or key. I wrote a small script that makes the same request for all 94 routes on the tracker, every minute or two, and saves the answers.

It ran from August 30 to September 30, 2026, and collected about 2.3 million bus positions from 575 different buses.

Then I compared those positions to Pace's published schedule (the GTFS feed Pace gives to Google Maps and other apps). For every scheduled trip, I checked each of its stops: did any bus on that route come within 500 meters of the stop within 15 minutes of the scheduled time? A trip where at least five stops could be checked, and not one of them was served, counts as a ghost trip: a trip on the schedule that never appeared on Pace's own tracker.

A few things this can and can't do:

• My computer wasn't watching all the time. The collector went down a few times (a reboot, a sleeping computer). Any stretch it wasn't watching is marked "unknown," not counted as a ghost.
• The tracker doesn't cover every route. Its route menu lists 94 routes. Another 40 routes that run on weekdays, about 17% of Pace's weekday trips, aren't on it at all, so I can't say anything about them. One more route, the 221 on Wolf Road, appears on the menu but showed just 2 positions all month. I left it out rather than call it a 100% ghost route.
• "Never appeared on the tracker" is not the same as "never ran." A bus that ran with its tracker switched off looks exactly like a bus that didn't run. I can't tell those apart, and nobody outside Pace can.
• The test is generous. Any bus on the route counts, in either direction, so a bus going the other way past a stop gives that stop credit. I'll come back to why that matters.

That left 123,187 scheduled trips on 86 routes that I could judge.

The whole system

Between 1.06% and 1.89% of the trips I could check never appeared on the tracker. That's roughly 1 in 94 to 1 in 53. On a typical weekday Pace schedules about 5,450 trips on these routes, and somewhere between 52 and 95 of them didn't show up.

It's a range because some trips are hard to call. When I looked closer at each ghost trip, some had a bus labeled with a different route passing a lot of their stops. Sometimes that's a bus running the trip under the wrong route number on the tracker; sometimes it's just two routes sharing a street. Trips where one such bus covered most of the stops I dropped as explained. The low end counts only trips where no single bus covered more than a fifth of the stops. The high end adds the ones in between, which I can't call either way.

under 1%1 to 3%3 to 6%6 to 10%10% or morenot on the tracker

All 86 measured routes as a table

Route	Name	Trips checked	Never appeared
366	Park Forest - Chicago Heights	1,069	10.8% to 26.4%
367	University Park - Park Forest	738	25.5% to 26.1%
357	Lincoln Highway	1,535	7.8% to 13.6%
356	Harvey - Homewood - Tinley Park	1,483	2.1% to 9.2%
542	Bluff City	1,171	3.0% to 9.1%
890	Chicago Heights - UPS Hodgkins Limited	232	3.0% to 9.1%
604	Wheeling - Schaumburg	1,005	8.5% to 8.7%
508	East Joliet	550	4.4% to 8.6%
564	Jackson / 14th	534	0.2% to 7.9%
354	Harvey - Oak Forest Loop	634	3.1% to 7.7%
358	Torrence	759	7.2%
509	Joliet-Fairmont	751	2.5% to 5.7%
471	Highland Park-Northbrook Court	413	0.0% to 5.6%
359	Robbins - South Kedzie Avenue	1,113	5.1% to 5.4%
310	West Madison - Hillside	807	0.1% to 5.2%
353	95th/Dan Ryan CTA-Calumet City-Homewood	1,761	4.9% to 5.1%
803	Carpentersville Local	876	3.1% to 4.8%
801	Elgin-Geneva	407	4.7%
350	Sibley	3,175	4.0% to 4.4%
549	South Randall	840	0.6% to 3.9%
392	Green Line Cicero CTA - UPS Hodgkins Limited	158	2.5% to 3.8%
317	Westchester	1,337	0.0% to 3.4%
395	95th/Dan Ryan CTA - UPS Hodgkins Limited	409	3.2%
314	Ridgeland Avenue	628	1.4% to 2.4%
223	Elk Grove - Rosemont CTA Station	2,669	0.7% to 2.2%
270	Milwaukee Avenue	784	0.0% to 2.2%
541	Northeast Elgin	1,192	0.2% to 2.1%
302	Ogden - Stanley	674	1.5% to 1.9%
472	Highland Park - Highwood	415	1.4% to 1.9%
561	Castlecrest via McAree	492	0.0% to 1.8%
563	Great Lakes Naval Station	630	0.2% to 1.6%
505	West Joliet Loop	1,073	0.3% to 1.6%
319	Grand Avenue	910	1.3% to 1.4%
571	Zion	1,148	1.1% to 1.3%
507	Plainfield	1,190	0.3% to 1.3%
308	Medical Center	2,051	0.1% to 1.2%
208	Golf Road	1,924	0.6% to 1.1%
316	Laramie Avenue	1,276	0.6% to 1.1%
305	East Roosevelt	1,382	0.9% to 1.1%
349	South Western	3,230	1.0% to 1.0%
215	Crawford-Howard	1,717	1.0%
552	North State - Spring Hill Mall	1,138	0.0% to 0.9%
290	Touhy Avenue	3,300	0.7% to 0.8%
543	Dundee-Carpentersville	787	0.1% to 0.8%
606	Rosemont - Schaumburg Limited	2,548	0.2% to 0.8%
524	West Aurora	563	0.4% to 0.7%
384	Narragansett - Ridgeland	573	0.0% to 0.7%
303	Forest Park - Rosemont	1,198	0.0% to 0.7%
332	River Road - York Road	761	0.0% to 0.7%
554	Elgin - Woodfield	763	0.4% to 0.7%
834	Joliet - Downers Grove	690	0.6%
574	CLC - Hawthorn Mall	575	0.5%
559	Illinois Route 59	1,041	0.0% to 0.5%
547	Wing Park	1,087	0.0% to 0.5%
548	Highland	1,089	0.0% to 0.5%
379	Midway - Orland Park	1,585	0.3% to 0.4%
272	Milwaukee Avenue North	974	0.4%
301	Roosevelt Road	1,977	0.2% to 0.4%
352	Halsted	4,567	0.3% to 0.4%
572	Washington	1,139	0.0% to 0.3%
570	Fox Lake - CLC	626	0.3%
364	159th Street	2,830	0.3%
100	Pulse Milwaukee Line	3,749	0.2%
330	Mannheim - LaGrange Roads	2,125	0.1% to 0.2%
568	Belvidere	1,386	0.2%
309	Lake Street	1,139	0.0% to 0.2%
331	Cumberland - 5th Avenue	1,879	0.1% to 0.2%
533	Northeast Aurora	1,270	0.2%
307	Harlem	2,958	0.1%
565	Grand Avenue	1,389	0.1% to 0.1%
381	95th Street	3,521	0.1%
250	Dempster Street	1,092	0.0% to 0.1%
313	St. Charles Rd.	1,212	0.0% to 0.1%
383	South Cicero	2,408	0.1%
386	South Harlem	1,235	0.1%
530	West Galena - Naperville	1,227	0.1%
501	West Jefferson	1,365	0.1%
315	Austin Boulevard	3,437	0.1%
322	Cermak Road - 22nd St.	3,481	0.0%
101	Pulse Dempster Line	3,888	0.0%
213	Green Bay Road	1,875	0.0%
311	Oak Park Avenue	1,631	0.0%
318	West North Avenue	3,207	0.0%
390	Midway CTA - UPS Hodgkins Limited	314	0.0%
722	Ogden Avenue	1,213	0.0%
802	Aurora - Geneva via Lake Street	1,233	0.0%

Where: the south suburbs

The missing trips are not spread evenly. The 10 routes with the most confirmed ghost trips account for 68% of them, while running only 11% of the trips.

These are the 10 routes with the highest rates, among routes with at least 100 trips checked:

Route	Name	Trips checked	Never appeared
366	Park Forest to Chicago Heights	1,069	11% to 26%
367	University Park to Park Forest	738	26%
357	Lincoln Highway	1,535	8% to 14%
356	Harvey, Homewood, Tinley Park	1,483	2% to 9%
542	Bluff City	1,171	3% to 9%
890	Chicago Heights to UPS Hodgkins	232	3% to 9%
604	Wheeling to Schaumburg	1,005	9%
508	East Joliet	550	4% to 8%
564	Jackson / 14th	534	0% to 8%
354	Harvey to Oak Forest	634	3% to 8%

Six of the ten are in the south suburbs, around Park Forest, Chicago Heights and Harvey. Route 604 in the northwest suburbs stands out because its range is narrow: about 9% either way. On Routes 366 and 367, about one scheduled trip in four never appeared.

I'm deliberately not saying anything here about who lives along these routes. That's worth asking, but it needs Census data, not a guess.

When: late nights and weekends

By time of day, the morning rush is the most reliable: under 1.3% of trips starting between 6 and 9am went missing. After 7pm it climbs, and trips starting around midnight go missing 6 to 7% of the time. Weekends are worse than weekdays, with Saturdays roughly twice the weekday rate.

Why my stop didn't show up in any of this

Here's the part that surprised me. By the whole-trip test, the 309 is one of Pace's most reliable routes: almost none of its trips are ghosts. If I'd stopped there, I'd have concluded my hour-long waits were bad luck.

The problem is the generous test. At rush hour the 309 runs every half hour in both directions. If one trip doesn't run, the buses before and after it, and the buses going the other way, still pass most of its stops within 15 minutes. The trip gets credit for stops it never served. On a busy route, a whole missing trip can look like it ran.

So that system-wide range is a floor. On frequent routes, the real number is probably higher.

To check my own stop properly, I stopped asking "did a bus come near the stop?" and asked the question a rider asks: did an eastbound bus actually leave Elmhurst? Elmhurst is the end of the line, so buses arrive, park for a few minutes and then leave. I counted a departure only when a bus that had been at the station was, within 15 minutes, at least a kilometer and a half away and heading east, and timed it at the moment it left.

Getting that right took a few tries, and two of my early versions were wrong. The first one timed each bus at the moment it arrived rather than when it left. Buses park about 200 meters up York Street before leaving, so a bus that sat there for ten minutes made it look like a 129-minute wait that never happened. I found that by tracing individual buses ping by ping, and only trusted the method once it matched those traces.

What happened at Elmhurst

I paired every scheduled eastbound departure in September with the real departures, one bus per scheduled time, allowing a bus to be up to 5 minutes early and up to 30 minutes late (or until the next scheduled bus, if that's sooner).

103 of 525 scheduled departures never left. That's about one in five. But it depends heavily on the time of day:

Weekdays	Scheduled departures checked	Never left
Before 6am	21	5%
Morning rush, 6 to 9am	126	25%
Midday, 9am to 2pm	100	4%
Afternoon rush, 2 to 7pm	161	34%
Evening, after 7pm	57	7%

At midday the 309 runs almost exactly to schedule. At rush hour, when the most people are trying to get home from the train, a quarter to a third of the scheduled buses never leave.

From the platform, that looks like this. Across all of September, the typical wait for the next bus was 4 minutes. But on weekday afternoons between 2 and 6pm, roughly a third or more of the scheduled departures were followed by a wait of over half an hour. On Wednesday, September 2, the 2:57, 3:33, 4:03, 4:33 and 5:03 buses never left. The next bus pulled out at 5:39pm, 2 hours and 42 minutes after the first one that didn't come. That's the kind of afternoon I'd been having.

It's the same trips, over and over. Some scheduled departures were missing on most of the weekdays I could check:

Scheduled departure	Weekdays missing
6:31am	12 of 21
2:57pm	9 of 17
5:03pm	9 of 17
3:33pm	8 of 18
6:01am	8 of 21
4:33pm	7 of 17

The 2:57 and the 5:03 are scheduled for the same bus.

The first trip of a bus's day is the one most likely to go missing. Pace's schedule data says which bus is meant to run each trip. When the Elmhurst departure is the first trip of that bus's day, it was missing 45% of the time (17 of 38, a small sample). When it's a later trip, 18%. That points less to buses running late in traffic and more to buses not getting out of the garage on time, or at all. Why that happens (drivers, vehicles, something else) is something only Pace can answer.

I checked the obvious alternative. Nearly every 309 bus also runs Route 313 during the day, switching at the Austin Boulevard end. If the tracker kept showing "313" on a bus that had switched to the 309, my method would have missed it. So I reran everything counting buses of any route label. Not one of the 103 missing departures turned out to have been run by a bus under another label.

What I can't tell you

• Whether a trip was cancelled or just very late. A bus more than 30 minutes late, or later than the next scheduled one, counts as missing. To someone standing at the stop, those feel the same.
• Whether a bus ran with its tracker off. I found one bus all month that left Elmhurst and then vanished from the tracker. Counting it or not changes nothing.
• Anything about the 40 routes the tracker doesn't cover.
• Whether September was typical. It's one month.

An earlier version

In early September I put up a live page for the Route 309, with a flyer at the stop, using a first version of this method. That version counted hours when my collector was switched off as missed buses, so it overstated the problem. It's still online, marked as out of date, and this page replaces it.