The significance of vestibular input for place navigation in the Morris water maze was examined in 11 hooded rats. A 5-min rotation (2 Hz) immediately preceding retrieval of an overtrained place navigation task prolonged escape latencies by about 10 s corresponding to circular swimming induced by postrotatory aftereffects, but did not otherwise interfere with target location. A 1-min rotation immediately following the acquisition trial in the working memory version of the water maze task did not deteriorate performance in the retrieval trial performed 5 min later. Two components of the acquisition trial, i.e., active search of the hidden target (up to 1 min) and latent learning during the 30-s stay on the platform, contribute almost equally to the formation of the working memory trace. A 1-min rotation immediately preceding the isolated platform learning component impaired subsequent retrieval but was ineffective when applied 3 min earlier. Latent learning was completely disrupted when the platform with the animal was rotated during the entire 30-s stay on the target platform or when the 30-s stay on the stationary platform was followed 0 or 3 min later by a 30-s rotation on the platform. The interfering effect of the latter procedure was suppressed by covering the platform with the animal by an opaque cylinder. It is concluded that vestibular cues are particularly important for the orientation of the animal in the gravitation field and for the estimation of the angles between vectors plotted from the animal toward external landmarks. Agreement between vestibular and visual signals is a prerequisite of efficient navigation.