This study investigates the impacts of geomagnetic storms on the performance of the International Reference Ionosphere 2016 (IRI-2016) foF2 predictions at different latitudes during geomagnetically disturbed periods in different seasons of the year 1989 which is around maximum solar activity. For this purpose, hourly foF2 data measured from the ionosonde stations Manila (14.7 degrees N, 121.1 degrees E), Rome (41.8 degrees N, 12.5 degrees E), Argentia NF (47.3 degrees N, 54.0 degrees W) and Uppsala (59.8 degrees N, 17.6 degrees E) that are located at low, middle and high latitudes, and hourly foF2 data calculated from the IRI-2016 for the same locations are used. In our study, a comparison between observational results and that of the IRI-2016 foF2 predictions is made. Planetary geomagnetic activity 3h-K-p and 3h-a(p) indices are used as geomagnetic activity indicators. In order to test the performance of the IRI-2016 for disturbed geomagnetic conditions, the percentile deviations and root mean square errors (RMSE) are calculated using the foF2 data from the IRI-2016 and the ionosonde stations. These analyzes are done for four seasons based on geomagnetic stormy days around equinoxes (March 21, September 23) and solstices (June 21, December 21). Therefore, it is found that during geomagnetically disturbed periods, the IRI-2016 foF2 predictions deviate considerably from the foF2 data taken from the ionosonde stations and this deviation differs according to the seasons and latitudes. The results show that the last version of the IRI model that includes the foF2 storm model is not sufficient to properly represent the real conditions of a disturbed ionosphere and there is still space for improving the IRI model for geomagnetically disturbed conditions.