BACKGROUND: Capturing neighborhood-specific physical activity is necessary for advancing understanding about the relations between neighborhood walkability and physical activity. This study examined the test-retest reliability of previously developed items (from the Neighborhood Physical Activity Questionnaire) for capturing setting-specific physical activity among Canadian adults. METHODS: Randomly sampled adults (n = 117) participated in two telephone-interviews 2 to 5 days apart. Respondents were asked a series of items capturing frequency and duration of transportation-related walking, recreational walking, moderate and vigorous-intensity physical activity undertaken inside and outside the neighborhood in a usual week. The test-test reliability of reported physical activity levels were then examined using Intraclass and Spearman's rank correlations, kappa coefficients, and overall agreement. RESULTS: Participation, frequency, and the duration of transportation-related and recreational walking and vigorous-intensity physical activity inside and outside the neighborhood showed moderate to excellent test-retest reliability. Moderate reliability was found for moderate-intensity physical activity undertaken inside (k = 0.48; ICC frequency = 0.38; ICC duration = 0.39) and outside (k = 0.51; ICC frequency = 0.79; ICC duration = 0.31) the neighborhood. CONCLUSIONS: Neighborhood-specific physical activity items administered by telephone-interview are reliable and are therefore appropriate for use in future studies that examining neighborhood walkability and physical activity.