BACKGROUND:  Walking is a popular recreational activity and a feasible travel mode.  Associations exist between walking and the built environment, but knowledge is lacking about specific environmental conditions associated with different purposes of walking.  METHODS:  This cross-sectional study used a survey of 438 adults and objective environmental measures. Multinomial logit models estimated the odds of walking for recreation or transportation purposes. RESULTS:  Utilitarian destinations were positively associated with transportation walking, but recreational destinations were not associated with any walking. Residential density was correlated with both purposes of walking, and sidewalks with recreation walking only. Hills were positively associated with recreation walking and negatively with transportation walking. CONCLUSIONS:  Physical environment contributed significantly to explain the probability of walking. However, different attributes of environment were related to transportation versus recreation walking, suggesting the need for multiple and targeted interventions to effectively support walking.