The problem of estimating parameters of multiple narrowband emitter signals from sensor array data is considered. Under the assumption of Gaussian distributed emitter signals, the stochastic maximum-likelihood (ML) technique is known to provide statistically efficient estimates, i.e., it achieves the Cramer-Rao bound (CRB). A multidimensional signal subspace method, termed weighted subspace fitting (WSF), has recently been proposed. It is shown that the WSF and ML estimates are asymptotically identical (for large data records). As a consequence, the WSF method is asymptotically efficient, assuming temporally white Gaussian signal waveforms and noise.