Durga Puja in Howrah Kolkata

Howrah is the city on the west bank of the Hooghly — a separate municipal corporation with its own pujas, its own crowds and the country's busiest railway station at the head of it.

Within it this site groups pujas into one neighbourhood: Howrah.

The river crossing shapes everything. Plan around the bridge and the ferry timetable rather than around the distance, which is short.

The practical unit of a Pujo evening is a neighbourhood, not a zone. Pick one cluster, walk it properly, and move only if you still have the legs for it.

All two have a confirmed position, so every one of them can go into a route, show a distance from where you are standing, and appear on the map.

Until more positions are confirmed, this zone is one to browse by name rather than to route through.

Tap Add to route on any puja below and it joins your route plan. The Route Planner orders your stops, draws the walk between the ones whose positions are confirmed, and keeps the list on this device between visits — so you can build the evening at home and follow it on the street.

Nothing on this page describes a 2026 theme, a timing or a crowd level, because no such thing has been published for these pujas. Where the site knows something it says it; where it does not, it leaves the line blank rather than filling it. A pandal listing with an invented opening time is worse than one with no opening time at all.

Names come from the PujoKolkata master dataset, which is the editorial source of truth for what these pujas are called, which neighbourhood they sit in and how they are organised. Coordinates come separately, from two sources: the IndianOil–Kolkata Police Puja Guide, and community-curated puja maps of the city. Where both know a puja the guide wins, and where a position is only roughly known the puja page says so. Nothing here is geocoded from an address or a postcode: a pin derived that way looks exactly as real as a true one while sending you to the wrong street.

If a puja you know is missing, or something here is wrong, it is a gap in the dataset rather than a judgement. The list is rebuilt from the source on every deploy, so a correction there appears here automatically.