We present a Bayesian approach for modeling unemployment duration when the data are subject to interval censoring and there is unobserved heterogeneity among individuals. The heterogeneity is accounted for by including a frailty term in the hazard function. We use a Weibull baseline hazard function and a gamma frailty distribution. The Bayesian inference is carried out using the Markov chain Monte Carlo (MCMC) method via WinBUGS software. The methodology is illustrated using simulated data and unemployment duration data from a longitudinal survey conducted in Iran.