Social events happening in a city can influence and affect a large number of the citizens, directly or indirectly involved. Having a metric to measure the popularity of such events can help in estimating the resources needed for handling them, improving facilities such as public transportation or traffic estimations. This paper reports on the problem of association of Tweet messages to social events in order to discover the popularity of an event. We propose and evaluate a method that computes an association coefficient for an event-tweet pair and discuss the results obtained.