Abstract Several hypotheses have been offered to explain the failure of many recent surveys to detect any adverse visitor-response to crowding at recreation sites. Surveys in the Forest of Dean have provided evidence for some of these hypotheses. Numerous interfering variables obscure the relationships between crowding and satisfaction that do exist. Broadleaved woodland is effective in reducing perception of crowding. Displacement of crowd-averse by crowd-tolerant visitors is an important phenomenon and can lead to bias in results. Use of a crowdaversion index improved satisfaction/density relationships. It is usually visitors without expectations about crowding who are insensitive to it. These results restore some credibility to the satisfaction/density model: forest managers should plan facilities with the adverse, but variable, effects of crowding in mind.