This paper contributes to the debate on identity politics by examining whether the religious composition of voters predicts electoral outcomes. Using assembly constituency-level data from six elections in West Bengal, India, between 2011 and 2024, we study how the Muslim population share relates to party performance. We show that religious composition becomes a much stronger correlate of electoral outcomes in the later period (from 2016 onwards): constituencies with higher Muslim shares increasingly align with the incumbent Trinamool Congress (TMC), while the Bharatiya Janata Party (BJP) consolidates support in constituencies with lower Muslim shares. We also document heterogeneity within predominantly Hindu constituencies. Economically vulnerable areas – proxied by higher shares of marginal agricultural labourers – remain relatively more supportive of the TMC, while better-off agricultural constituencies shift towards the BJP. Together, the results suggest that West Bengal’s recent electoral realignment reflects both strengthening religious polarisation and an interaction between identity-based mobilisation and material considerations, with implications for political competition and accountability in democracies.