In order to assess the reliability of a building energy simulation program (TRNSYS) from the standpoint of user, a set of verification experiment and calculation of cooling load for a test space is carried out. This work is a complement of the previous study that dealt with heating load for the same space. The test space is kept airtight to eliminate the source of uncertainties in modeling. A window-mounted, on/off controlled air-conditioner is used for cooling, whose performance has been established a priori. The calculation encompasses two models for evaluating cooling load in TRNSYS: energy rate control and temperature level control. Comparison of the total cooling loads obtained from different sets of experimental data enables to validate the measurements. The experimental result shows that the latent load is fairly large even in the absence of apparent air change in the space, which needs to be clarified. Each of hourly and daily accumulated sensible loads is compared between the experiment and two calculation models. Despite an inconsistency associated with solar irradiation, both of the models agree favorably with the experiment within a tolerance, illustrating their capability of properly predicting space thermal loads.