Since 2002, human cases of West Nile virus (WNV) have occurred every year in southern Canada, but WNV risk remains challenging to predict. Here, we explored the ability of weather-based forecasting models to predict the seasonal abundance of two WNV vector species (Culex pipiens-restuans and Aedes vexans) in Québec, Canada, and explored the importance of accounting for larvicide use and local habitat (forest park vs residential garden). A gamma-generalized linear model predicting mosquito abundance was developed based on an approach previously used in Ontario combining temperature and precipitation during the days preceding mosquito captures. This model was calibrated and validated for each species with independent entomological datasets from the Montréal region collected in 2013 and 2014. Culex pipiens-restuans abundance was associated with mean degree days (dd; >9°C) over the 22 d before mosquito capture and with mean precipitation over the 71 d before capture; Ae. vexans abundance with the mean dd (>12°C) over the 24 d before capture and mean precipitation over the 30 d before capture. These results are consistent with temperature effects on immature development rates and adult activity, and effects of precipitation on the abundance and suitability of breeding sites. Taking into account larvicide use and habitat significantly improved the predictions. This study provides evidence that weather conditions can yield robust short-term predictions of the regional daily mosquito abundance, particularly when accounting for local variation in habitat or mosquito control efforts, and may provide real-time indicators of WNV or other mosquito-borne disease risks during the summer.