The annual anomaly is the ionospheric phenomena in which the globally-averaged electron density is greater in January than it is in July. This anomaly causes the ionospheric solsticial variation – a variation with a periodicity of one year that is in-phase with the January solstice – to be more pronounced over the Northern Hemisphere than the Southern Hemisphere. Predictions of the magnitude of annual anomaly using the International Reference Ionosphere (IRI) model have been shown to be unreliable so far. The objective of our study is to investigate model prediction of the magnitude of the annual ionospheric anomaly using new ionospheric indices as inputs in the IRI model. These new indices improve predictions ionospheric variations that differ over the two hemispheres. We present a retrospective analysis of the IRI predictions of the ionospheric daytime annual anomaly and solsticial variation using a model-data comparison with observations from over 40 ionosondes for high, moderate, and low solar cycle conditions. Our results show that there is an overall 33 % underestimation of the magnitude of the annual anomaly when the by the IRI. When the new ionospheric indices as used in the IRI, model predictions underestimate the magnitude of the annual anomaly by 6 %. This indicates an improvement of the model predictions when using the new indices. We show that the underestimation of the annual anomaly by IRI is related to a similar underestimation of the magnitude of the ionospheric solsticial variation over the Northern Hemisphere. Based on our results, we infer that the underlying processes of the annual anomaly must vary across each hemisphere.