Direct current (DC) link electrolytic capacitors are one of the components most prone to fail in power electronics converters. The traditional capacitor condition monitoring methods require extra hardware, which translates as an increased cost. This paper proposes and experimentally evaluates an alternative software-based condition monitoring method that uses an artificial neural network (ANN) to predict the capacitance of the DC-link capacitor bank in three-phase front-end diode rectifier motor drives. Based on time-domain parameters, the ANN is trained with a printed circuit board capacitor jig and evaluated with aged samples. Experiments were conducted in several operating conditions and the absolute prediction errors were all less than 2.4%, showing that the proposed method is able to monitor the degradation level of the dc-link capacitor bank in variable frequency drives.