Objective. Increased GSK3B activity has been reported as a state marker of major affective episodes in patients with depression and bipolar disorder. No study so far has addressed GSK3B activity in late-life depression. The aims of the present study were to determine GSK3B activity in platelets of elderly patients with major depression, and the association between GSK3B activity and the severity of depressive symptoms and cognitive impairment. Methods. Forty drug-free elderly patients with major depressive episode were compared to healthy older adults (n = 13). Severity of the depressive episode and current cognitive state were determined by the Hamilton Depression Scale (HAM-D) and the Cambridge Cognitive Test (CAMCOG), respectively. Total- and ser-9-phosphorylated GSK3B (tGSK3B and pGSK3B) were determined in platelets by enzyme immunometric assays (EIA). GSK3B activity was indirectly inferred by the GSK3B ratio (i.e. pGSK3B/tGSK3B). Results. Elderly depressed patients had significantly lower pGSK3B levels (P = 0.03) and GSK3B ratio (P = 0.03), indicating higher GSK3B activity. Higher GSK3B activity were observed in patients with severe depressive episode (HAM-D scores >22, P = 0.03) and with cognitive impairment (CAMCOG scores <86, P = 0.01). Conclusion. The present findings provide additional evidence of the involvement of GSK3B in the pathophysiology of late-life major depression. Higher GSK3B activity may be more relevant in those patients with more severe depressive symptoms and cognitive impairment.