Given an independent and identically distributed sample of angles from some absolutely continuous circular random variable with unknown probability density function \( f \), in this paper we address the estimation of the rth derivative of \( f \) by considering a class of higher-order kernel estimators which is inspired in the corresponding class of kernel estimators for data on the real line. It is proved that the estimators of the proposed class are asymptotically unbiased and weakly pointwise consistent, and an asymptotic expansion for its mean integrated squared error is given. A finite sample study indicates that higher-order kernels, asymptotically improving the rate of convergence, may be in general useful for estimating the rth derivative of \( f \) when moderate or large samples are available to the user.