Background: Calcium (Ca) and magnesium (Mg) are crucial in metabolism, excitability and neuroglial plasticity. Our aim was to evaluate whether Mg (20 mg/kg) or Ca (100 mg/kg) could improve the memory prognosis in the kainic model of mesial temporal epilepsy. Methods: Seizures were induced by systemic injection of kainate (8mg/kg) and mice were then treated by ions every 48 hours. A placebo (physiological solution) replaced kainate or ions in specific groups. Six cohorts were studied for seven weeks: control group (G0: no kainate and no ion, only placebo); untreated reference group (GR: kainate and then placebo); G1 groups were treated from the third day (G1m, G1c: kainate and then Mg/Ca); G2 groups were treated from the third week (G2m, G2c: kainate and then Mg/Ca). Radial maze and a classic maze were used for cognition evaluation. Results: The memory (short/long term) was differently affected by kainate or improved by Mg/Ca. The treated groups performed better than GR mice, but Mg was more effective. In addition, Mg demonstrated an increasing therapeutic effect over time while Ca showed an acute and apparently decreasing action in the G1c group. Conclusions: Mg should be considered for a clinical evaluation of its effect on epileptic disorders.