The R package `npcure` implements a completely nonparametric approach for estimating the cure probability and latency functions in mixture cure models, considering right-censored survival times. The package provides nonparametric estimators for the cure probability and latency as functions of covariates, along with bootstrap bandwidth selectors for these estimators. It also includes a nonparametric covariate significance test for the cure probability, applicable to continuous, discrete, or qualitative covariates. The package addresses the challenge of developing procedures for testing the parametric form of the conditional survival function of uncured individuals with covariates, which is particularly difficult due to the curse of dimensionality. The nonparametric estimators are consistent under a specific condition that ensures the absence of cured individuals surviving beyond the largest possible censoring time. The package includes functions for estimating the cure rate and latency, selecting bandwidths for these estimators, and performing covariate significance tests. An example using bone marrow transplantation data illustrates the application of the package's functions.The R package `npcure` implements a completely nonparametric approach for estimating the cure probability and latency functions in mixture cure models, considering right-censored survival times. The package provides nonparametric estimators for the cure probability and latency as functions of covariates, along with bootstrap bandwidth selectors for these estimators. It also includes a nonparametric covariate significance test for the cure probability, applicable to continuous, discrete, or qualitative covariates. The package addresses the challenge of developing procedures for testing the parametric form of the conditional survival function of uncured individuals with covariates, which is particularly difficult due to the curse of dimensionality. The nonparametric estimators are consistent under a specific condition that ensures the absence of cured individuals surviving beyond the largest possible censoring time. The package includes functions for estimating the cure rate and latency, selecting bandwidths for these estimators, and performing covariate significance tests. An example using bone marrow transplantation data illustrates the application of the package's functions.