IMPORTANT NOTICE:
remtracker.ca is an independent, unofficial project. It is not affiliated with, endorsed by, sponsored by, or operated by CDPQ Infra, the Réseau express métropolitain (REM), or any related entity.remtracker.ca is an early-stage, work-in-progress project that aims to independently track, quantify, and visualize the real-world reliability of Montreal's REM. This website is just getting started, and many features, datasets, and improvements are actively under development. Over time, users can expect the following additions and enhancements:
The data currently presented on remtracker.ca covers the period from November 17, 2025 to January 24, 2026 only. This time range is intentionally highlighted, as no claims are made about completeness or coverage outside of these dates at this stage of the project.
The dataset was obtained through extensive and creative, non-traditional data engineering methods applied to publicly available information. These methods were used to reconstruct historical service interruptions and slowdowns after the fact, based on how, when, and where service notifications were publicly communicated.
As a result, the accuracy of the data is inherently dependent on the accuracy, timing, and completeness of the REM’s own historical public service notifications. In other words, the dataset reflects what was publicly communicated at the time, rather than an independent ground-truth measurement of operational events.
The data collection and reconstruction process is actively being refined, and the quality, resolution, and internal consistency of the dataset are expected to improve over time as methods are adjusted, validated, and expanded. Earlier periods and finer-grained detail will be added progressively as confidence in the methodology increases.
Each day is treated as a single “planned service window” in Montréal local time, and two totals are computed over that window: unique slowdown minutes and unique interruption minutes.
For a given date, REM messages (alerts and updates) are reviewed, and only those indicating a service state change are retained: slowdown start, interruption start, “gradual resumption” (interpreted as the end of an interruption and the start of a slowdown), and “service normal” (end of any disruption). General announcements, explanations, and replies are ignored unless they clearly signal a change in service state.
A timeline of service states is then reconstructed over the planned service intervals. Minutes are counted only within planned service time (any time outside the planned interval is discarded). Disruptions are measured using unique minutes, meaning that each calendar minute within the service window is counted at most once, regardless of how many messages reference it or how many network segments are affected during that minute.
For example, if an unplanned interruption affects only a portion of the network from 06:00 to 07:00, this still results in 60 interruption minutes being counted for the day. The analysis currently treats the REM as a single system-wide entity: a minute is considered disrupted if any part of the network is disrupted during that minute. Overlapping or repeated messages are therefore deduplicated by taking the union of time coverage rather than summing individual alerts. In future iterations, the analysis may become more granular and distinguish between individual network segments, but the current scope focuses on the network as a whole.
Severity rule: if, during the same minute, any part of the network is in interruption while another part is only slowed, that minute is counted as an interruption (the worst outcome prevails). Slowdown minutes are counted only for minutes where slowdown is present and no interruption exists anywhere on the network.
Cross-midnight rule: planned service windows may extend past midnight (for example, 05:30–01:10). In such cases, times after midnight up to the end of service (and generally up to 05:29) are treated as belonging to the previous service day’s window, so disruptions occurring at 00:xx are attributed to the previous date.
When no explicit “end” message is published (no “service normal” or “gradual resumption”), the disruption is assumed to persist until the end of the planned service window for that date. The final output for each day consists solely of the total slowdown minutes and total interruption minutes, uniquely counted and clamped to planned service time.
remtracker.ca is an independent, unofficial, third-party project created for informational and analytical purposes and, honestly, because the REM made Bashar Eskandar mad after it caused him to waste hours of sleep twice in the same week. On both occasions, he had to wake up early for classes at McGill University, only for the REM service to be interrupted, forcing him to sit in the cold inside a shut-down train for nearly an hour before eventually going back home and missing his classes anyway. This website is not affiliated with, endorsed by, sponsored by, or operated by CDPQ Infra, the Réseau express métropolitain (REM), or any of their subsidiaries, partners, contractors, or service providers.
All information presented on this site is derived from publicly available data sources and independent analysis. remtracker.ca does not represent an official service communication channel, operational control system, or real-time dispatch authority. Users should always rely on official REM communication channels for authoritative service status, traveler guidance, and operational updates.
While reasonable efforts are made to ensure clarity and internal consistency, the information presented on this site may be incomplete, delayed, approximate, or subject to interpretation. No guarantees are made regarding accuracy, completeness, or suitability for decision-making.
“REM” and “Réseau express métropolitain” are names, trademarks, or identifiers associated with CDPQ Infra. Their use on this website is purely descriptive, for the sole purpose of identifying the transportation system being analyzed, and does not imply any official relationship, endorsement, or authorization.
The underlying sources referenced by this project are publicly available. However, the dataset and visualizations on remtracker.ca reflect many hours of manual and automated work: collecting, cleaning, parsing, deduplicating, time-aligning, interpreting ambiguous messages, and consolidating everything into a consistent historical record.
Please do not copy, re-host, republish, or sell remtracker.ca’s compiled dataset or derived outputs as-is (or close variants) without permission. If you want to use it for research, journalism, or a community project, reach out and/or credit remtracker.ca clearly.
Unless explicitly stated otherwise, all original site content (text, charts, code, and the compiled dataset structure) is provided under terms that require attribution and prohibit redistribution of the dataset as a substitute for this site.