A　partially　linear　model　with　longitudinal　data　is　considered,　empirical　likelihood　to　inference　for　the　regression　coefficients　and　the　baseline　function　is　investigated,　the　empirical　log-likelihood　ratios　is　proven　to　be　asymptotically　chi-squared,　and　the　corresponding　confidence　regions　for　the　parameters　of　interest　are　then　constructed.　Also　by　the　empirical　likelihood　ratio　functions,　we　can　obtain　the　maximum　empirical　likelihood　estimates　of　the　regression　coefficients　and　the　baseline　function,　and　prove　the　asymptotic　normality.　The　numerical　results　are　conducted　to　compare　the　performance　of　the　empirical　likelihood　and　the　normal　approximation-based　method,　and　a　real　example　is　analysed.