Summary We explored the relationship of aquatic midge assemblages with physicochemical and biological environmental gradients to assess which variables may govern the distribution of midge larvae in 47 shallow, polymictic lakes across New Jersey and New York State (NJ/NY) in the United States of America. Subfossil taxa collected from surficial sediments (0–1 cm sediment depth), comprising 50 taxa from 47 lakes, were analysed in conjunction with environmental variables using multivariate statistical techniques. Alkalinity, maximum depth, surface area, total phosphorus and pH were identified as significant and ecologically relevant variables that explained the most variation in the midge assemblage across NJ/NY. Lake trophic state was the main driver for midge distributions in NJ/NY lakes. Biological gradients, such as per cent macrophyte cover or algal productivity (as chlorophyll a concentration), did not explain a significant portion of the variation in midge community composition. The addition of chaoborid larvae to ordinations strengthened the relationship between midge community structure and bottom oxygen concentration in NJ/NY lakes. Our results confirm that complex species–environment relationships in shallow, polymictic lakes create challenges for assessing midge assemblages along a particular environmental gradient, independently of other environmental conditions. However, it may still be possible to develop palaeolimnological inference models using midge remains to assess general historical patterns of disturbance for NJ/NY and other polymictic lakes, provided it is understood that midge‐based inferences integrate some covariation in changes along several environmental gradients.