Bipolar disorders still represent a global unmet medical need and pose a requirement for novel effective treatments. In this respect, glycogen synthase kinase 3β (GSK-3β) aberrant activity has been linked to the pathophysiology of several disease conditions, including mood disorders. Therefore, the development of GSK-3β inhibitors with good <i>in vivo</i> efficacy and safety profile associated with high brain exposure is required. Accordingly, we have previously reported the selective indazole-based GSK-3 inhibitor <b>1</b>, which showed excellent efficacy in a mouse model of mania. Despite the favorable preclinical profile, analog <b>1</b> suffered from activity at the hERG ion channel, which prevented its further progression. Herein, we describe our strategy to improve this off-target liability through modulation of physicochemical properties, such as lipophilicity and basicity. These efforts led to the potent inhibitor <b>14</b>, which possessed reduced hERG affinity, promising <i>in vitro</i> ADME properties, and was very effective in a mood stabilizer <i>in vivo</i> model.