Fishery managers use data on catch and harvest rates collected with both on-site and off-site angler surveys. Many researchers have hypothesized that catch rates from these surveys will differ systematically due to various biases. However, few direct comparisons have been made between the two survey methods using the same waters and seasons. We compared catch and harvest rates for three coldwater species, individually and combined, from concurrent mail and on-site surveys on five Maine lakes during the 1994 ice fishing season. Comparisons were also made for two Maine lakes during the 1994 open-water season. Estimates of catch and harvest from the mail survey exceeded the rates from on-site surveys in 28 of 38 comparisons for individual species. The average on-site to mail survey ratio was 0.40 for catch rates and 0.66 for harvest rates but varied significantly by lake, season, and species. The results suggest that mail surveys do not produce accurate estimate of catch and harvest rates. In situations where only mail survey data are available, fisheries managers should recognize that mail survey catch and harvest rates are likely to be overestimates compared with on-site data.