Data fusion is the study of optimal information processing in distributed multisensor environments with the benefits of more reliable detection and higher immunity to noise and sensor failures. We investigate applications of the data fusion principle to the handoff problem in popular code-division multiple access (CDMA) cellular wireless communication networks. When a mobile station traverses a cell boundary, cell handoff is required to make the transition of a mobile between cells smoother without degrading performance. CDMA multiple access allows soft handoff, in which the signal strength of the mobile is detected and tracked by the searchers from more than one base station. The cell receiving the strongest signal strength handles signal processing. We demonstrate that the signal detection quality in the handoff region can be further improved by fusing the detected results from multiple base stations. A new fusion handoff strategy is designed, and its error performance measures are derived analytically in the presence of log-normal shadowing. Theoretical fading margin analysis shows that the fusion handoff approach offers better network performance over conventional hard and soft handoffs.