F1Oostveer
To Zamenhofstraat101.2%
To Azartplein101.1%
Limited data50% comparable
How often does the ferry run?
The timetable compared with what we measure on the IJ.
Choose a period
5 Oct – 1 Nov 2026
CSVIJ ferries
0%20%40%60%80%100%120%
F2IJpleinveer
To IJplein97.7%
To Centraal Station97.8%
98% of scheduled time comparable
F3Buiksloterwegveer
To Buiksloterweg76.5%
To Centraal Station76.5%
99% of scheduled time comparable
F4NDSM-veer
To NDSM-werf106.6%
To Centraal Station106.6%
96% of scheduled time comparable
F6Distelwegveer
To Distelweg100.2%
To Pontsteiger100.2%
97% of scheduled time comparable
F7NDSM-veer · Pontsteiger
To NDSM-werf97.2%
To Pontsteiger99.6%
96% of scheduled time comparable
F9Sporenburgveer
To Zeeburgereiland98.4%
To Sporenburg96.6%
97% of scheduled time comparable
100% = the same average frequency as planned during comparable periods. This is not the percentage of departures on time. Choose a route to explore its patterns.
Also the North Sea Canal ferriesF20 · F21 · F22
F20Hempont
To Zaandam98.1%
To Hempontplein97.2%
99% of scheduled time comparable
F21Buitenhuizerpont
To Assendelft95.1%
To Spaarndam94.2%
95% of scheduled time comparable
F22Velsenpont
To Velsen Noord95.9%
To Velsen Zuid93.7%
98% of scheduled time comparable
How we measure
We compare how often the published timetable says a ferry should run with the intervals between observed crossings, for each line and direction.
- The timetable
- We use GVB’s timetable from OVapi’s national GTFS feed. We extract the planned frequency for the date being measured. It can change throughout the day. Each observation is compared with the plan that applies at that time.
- The crossings
- AIS messages report vessel positions. Repeated movements between the two shores help us assign a ferry to a line. Only passages with enough evidence for that line are included.
- The comparison
- 100% means the ferries ran as often as planned on average during comparable, measurable periods. It does not tell us whether every departure was on time, how long passengers waited or how many sailings were cancelled.
Calculation and limitations
Both directions in the pattern grid
The pattern grid pools observations from both directions. We divide the total observed intervals by the total expected from the timetable over exactly the same measurable time. Periods and directions with more expected intervals therefore carry more weight; we do not take a simple average of percentages.
If the timetable changes within an interval, each part is calculated using the frequency that applies there. The timeline keeps the two directions separate. A measured passage is not a measured departure from the quay.
What we cannot see
Missing AIS messages can make an interval appear longer. Insufficient observations, an unknown timetable and a period with no scheduled service therefore each have a distinct label. Missing data is never automatically treated as a missed sailing.
Measurement coverage shows how much of the scheduled service time is comparable. It does not prove that every ferry was observed. Reception coverage for imported historical AIS data is unknown.
Intervals longer than 60 minutes are excluded from this frequency calculation. These figures therefore do not provide a complete account of cancellations or the longest interruptions.
Sources: GVB timetable via OVapi (GTFS) · Open Waters AIS. All times are shown in Amsterdam time.
For the love of ferries
I’m Elmar. I love the ferries, but found it frustrating that public live data was unavailable. I wanted to check on my phone: am I missing the boat, or is it not running? That’s why I started this project.
With pont.fyi, I want to show how often the ferries run compared with the timetable. Differences from that timetable are not a criticism of the skippers who take us across every day.