Habitat rehabilitation efforts are predicated on the frequently untested assumption that habitat is limiting to populations. These efforts are typically costly and will be ineffective if habitat is not limiting. Therefore it is important to assess, rather than assume, habitat limitation wherever habitat rehabilitation projects are considered. Catch-count data from a standardized probability-based stratified-random monitoring programme were examined for indirect evidence of backwater habitat limitation by centrarchid fishes in the Upper Mississippi River System. The monitoring design enabled fitting statistical models of the association between mean catch at the spatial scale of tens of river kilometres and the percentage of contiguous aquatic area in backwater at least 1 m deep by maximizing a stratum-area weighted negative binomial log-likelihood function. Statistical models containing effects for backwater limitation failed to account for substantial variation in the data. However, 95% confidence intervals on the backwater parameter estimates excluded zero, indicating that population abundance may be limited by backwater prevalence where backwaters are extremely scarce. The combined results indicate, at most, a weak signal of backwater limitation where backwaters are extremely scarce in the lower reaches, but not elsewhere in the Upper Mississippi River System. This suggests that habitat restoration projects designed to increase the area of backwaters suitable for winter survival of centrarchids are unlikely to produce measurable benefits over intermediate spatial scales in much of the Upper Mississippi River System, and indicates the importance of correct identification of limiting processes. Published in 2004 by John Wiley and Sons, Ltd.