We describe Monte Carlo simulation results for breast tumor imaging using a compact, discrete gamma camera. The simulations were designed to analyze and optimize camera design, particularly collimator configuration and detector pixel size. Simulated planar images of 5-15 mm diameter tumors in a phantom patient (including a breast, torso, and heart) were generated for imaging distances of 5-55 mm, pixel sizes of 2x2-4x4 mm2, and hexagonal and square hole collimators with sensitivities from 4000 to 16,000 counts/mCi/sec. Other factors considered included T/B (tumor-to-background tissue uptake ratio) and detector energy resolution. Image properties were quantified by computing the observed tumor fwhm (fullwidth at half-maximum) and S/N (sum of detected tumor events divided by the statistical noise). Results suggest that hexagonal and square hole collimators perform comparably, that higher sensitivity collimators provide higher tumor SM with little increase in the observed tumor fwhm, that smaller pixels only slightly improve tumor fwhm and S/N, and that improved detector energy resolution has little impact on either the observed tumor fwhm or the observed tumor S/N.