In this talk we consider partially linear varying coefficient models.We provide the robust orthogonality-based estimator of the the parametric part as well asthe jump-preserving(JP)estimator of the nonparametric part.Specifically,we firstpresent an orthogonality-based estimation(OBE)method,a JP fitting procedure and alogarithm kernel smoothing technique for estimating the parametric part,nonparametric part and variance of the error term,respectively.The resultingestimators are proved to be asymptotic normal and consistent,respectively.Excitedly,the JP estimator not only gives smooth estimates of the continuity part of thecoefficient functions,but also maintains the desirable properties of the local linear35 estimator with regard to the bias and the boundary estimation while it estimates thejumps consistently,i.e.,it preserves the jumps well.Then,by applying the local linearsmoothing method and taking the estimated error heteroscedasticity into account,wesuggest the reweighted estimations of the parametric and nonparametric parts.Theresulting reweighted estimator of parametric part is shown to have smaller asymptoticvariance than the OBE estimator that neglects the error heteroscedasticity whileremaining the same bias,and the reweighted JP(RJP)estimator shares the samedistribution and advantages with the JP estimator.Moreover,several simulationexamples are presented to evaluate the finite sample performance of the proposedmethodologies.Finally,an application with financial data illustrates the usefulness ofthe proposed techniques.