Most theories of the programming of saccadic eye movements (SEM) agree that direction and amplitude are the two basic dimensions that are under control when an intended movement is planned. But they disagree over whether these two basic parameters are specified separately or in conjunction. We measured saccadic reaction time (SRT) in a situation where information about amplitude and direction of the required movement became available at different moments in time. The delivery of information about either direction or amplitude prior to another reduced duration of SRT demonstrated that direction and amplitude were specified separately rather than in conjunction or in a fixed serial order. All changes in SRT were quantitatively explained by a simple growth-process (accumulator) model according to which a movement starts when two separate neural activities, embodying the direction and amplitude programming, have both reached a constant threshold level of activity. Although, in isolation, the amplitude programming was faster than the direction programming, the situation reversed when two dimensions had to be specified at the same time. We conclude that beside the motor maps representing the desired final position of the eye or a fixed movement vector, another processing stage is required in which the basic parameters of SEM, direction and amplitude, are clearly separable.