In suitable robotic applications, throwing an object instead of placing it has the potential of improving the cycle time. In this context, a challenge is to generate time-optimal Pick-and-Throw (P&T) trajectories in order to further increase productivity. This paper introduces a methodology to determine a minimum-time throwing motion. This methodology consists essentially in determining an optimal release configuration (i.e. position and velocity) allowing an object to be thrown towards a desired target while minimizing the travel time of the throwing motion of the robot. To validate the potential of the proposed P&T approach, a comparison with the standard Pick-and-Place (P&P) process and an existing P&T method is made using the Delta-like parallel robot T3KR under different operating conditions. The obtained experimental results demonstrate the superiority and efficiency of the proposed P&T approach over the usual P&P and the existing P&T methods in terms of picking speed and cycle time.