Nuclear magnetic resonance (NMR) is a powerful technique for revealing biomolecular structure and dynamics. Technical advances over the past decade, including cryogenic probes, higher magnetic fields, and novel pulse sequences, allow for the analysis of increasingly larger and more complex systems. Multidimensional, multinuclear experiments are invariably required to resolve individual resonances. However, the increased sampling rate imposed by the Nyquist theorem at higher magnetic fields (due to greater spectral dispersion) means that experiment times become prohibitively long when conventional uniform sampling is employed. Non-Fourier methods of spectrum analysis open the possibility of nonuniform sampling, permitting data to be collected at short evolution times to ensure sensitivity and at long evolution times to allow for high resolution. The combination of nonuniform sampling methods with non-Fourier methods of spectrum analysis enable the computation of high resolution multidimensional spectra from much shorter data records than can be employed using conventional Fourier and uniform sampling methods. The design of optimal sampling strategies, for minimizing sampling time while maintaining resolution and sensitivity, or optimizing sensitivity and/or resolution for a given total experiment time remains an open challenge. In the present work the maximum entropy (MaxEnt) method of spectrum reconstruction is used to characterize the performance of nonuniform sampling strategies with respect to sensitivity and resolution. A peak identification algorithm is developed along with a metric for spectra comparison. These tools are used to investigate whether prior knowledge of peak frequencies can aid the design of optimal sampling strategies. Such strategies could be useful in a number of contexts in biomolecular NMR, including relaxation studies and multidimensional experiments conducted subsequent to “scout” experiments, such as two-dimensional HSQC.