PURPOSE: Unilateral dystonic limb posturing in partial seizures has been shown to be an accurate lateralizing sign indicating seizure onset in the contralateral hemisphere. However, its clinical utility may be reduced by confusion with other lateralized ictal motor phenomena. In this study, the ictal phenomena of dystonic limb posturing, tonic limb posturing, unilateral immobile limb, and version were distinguished and examined in patients with temporal and extratemporal seizures. METHODS: Partial seizures in 54 patients, successfully treated by surgery (34 temporal, 20 extratemporal; 14 frontal, 3 parietal, and 3 occipital), were analyzed blindly by 3 reviewers. Interobserver agreement was tested with kappa indexes and positive predictive value (PPV) was determined for each sign. RESULTS: In patients with temporal lobe epilepsy (TLE), dystonic posturing occurred in 35.3% (kappa 0.78, positive predictive value (PPV) for the sign being contralateral to seizure onset 92%); tonic limb posturing occurred in 17.7% (kappa 0.36, PPV 40%); unilateral immobile limb occurred in 11.8% (kappa 0.23, PPV 100%); and version occurred in 35.3% (kappa 0.77, PPV 100%). In patients with extratemporal epilepsy, dystonic posturing occurred in 20.0% (kappa 0.31, PPV 100%); tonic limb posturing occurred in 15.0% (kappa 0.08, PPV 67%); and version occurred in 40.0% (kappa 0.54, PPV 100%). The higher kappa indexes were significant for dystonic posturing (p < 0.001) and tonic limb posturing (p = 0.032) in TLE. Dystonic posturing (p = 0.034), tonic posturing (p = 0.07), and version (p = 0.0038) occurred earlier in extratemporal seizures than in temporal seizures. CONCLUSIONS: Of the limb ictal motor phenomena, only dystonic posturing was accurate and had good interobserver agreement.