We present asymptotic distribution theory for the inverse-density-weighted quasi-maximum likelihood estimator of semi-parametric index models proposed by Ruud. We also compare the performance of this estimator with the average derivative estimators proposed by Stoker.