Methodology is developed to estimate attendance at day-long, non-ticketed, non-gated events. It consists of two complementary surveys. First, instantaneous counts of individuals are recorded at one or more times during the event. Second, either an access-survey, where parties who are leaving the event are intercepted, or a roving-survey where parties are intercepted during attendance at the event, record information on the start and end of the attendance for that party. The proportion of parties that were active during the instantaneous counts is used to expand the instantaneous counts. The estimators account for the cluster sampling of individuals within parties and the length-biased sampling in the roving survey. The methodology is applied to estimating attendance at a community festival in Vancouver, Canada.