The nominal depth resolution achieved in confocal Raman microscopy is on the order of a few micrometers. Often, however, the depth resolution is decreased by light refraction at the sample surface. The problem can be avoided with the use of an immersion objective and index matching oils. Through this intervention the instrument point-spread function (PSF) can be assumed to be independent of the depth of focus in the sample, and spatially invariant depth profiles can be acquired. In this work the instrument PSF was determined by measuring a depth profile of a thick uniform sample and calculating the first derivative of the depth profile curve. The first-derivative method was also used to determine sample thickness. Convolution with the PSF makes it possible to simulate the behavior of the instrument with different sample functions. It is also possible to use the instrument PSF to deconvolve depth-profiling data. Deconvolution reduces the blurring effect of the instrument and increases the depth resolution. Deconvolution can also be used in analysis of the sample surface position and in layer structure analysis. In this paper we show how the convolution integral can be used with the immersion sampling technique to determine the PSF and how the sample thickness can be determined.