We investigate the estimation of amplitude of minimum shift keying in the presence of co-channel interference. We consider the maximum likelihood (ML) estimation technique for amplitude estimation, for the practical case of sampled waveforms. We assume a single interfering user, and derive the full ML estimator as a function of the number of samples per symbol, and the sampling phase. We review analysis for corresponding rectangular pulse cases for insight, and also consider the effect of both bit and carrier phase synchronism on performance. Using high signal-to-noise ratio approximations, we show that the ML estimator for MSK becomes a scaled version of that used for rectangular pulses, with the scaling dependent upon the sampling rate and sampling phase. The accuracy of the amplitude estimator was verified by performing a single stage of interference cancellation on a practical system consisting of two users in the presence of Additive White Gaussian Noise. Numerical results are included to corroborate the analysis.