High-resolution isophotal light profiles have been determined for 36 field galaxies in images taken for the Hubble Space Telescope Medium Deep Survey prior to HST's refurbishment mission. The images were obtained with the Wide Field Camera in "parallel mode" and deconvolved using the Lucy-Richardson image-restoration algorithm with a large library of observed stars as PSFs. From ground-based spectroscopy, redshifts have been determined for 32 of the galaxies, yielding z=0.026-0.554. The restored images, light profiles, and spectra of individual galaxies are discussed. By fitting the light profiles with models of exponential disks and/or r¼ bulges, values for scale lengths (rs and/or re) and bulge-to-disk ratios have been derived. Our main results are: (1) The measured half-light radii, rhl, are in agreement with those obtained independently via multiparameter fitting of the raw data, validating recent tests of galaxy evolution models based on these angular-size determinations. However, since some redshifted objects that are clearly extended in the deconvolved images have been described as stellar using fits to the aberrated data, application of the latter technique may result in a significant fraction of compact galaxies being misclassified as stars. Understanding this bias towards errant stellar classifications is of considerable importance when using automated fitting algorithms to address the contribution of compact objects to faint galaxy counts. (2) The distribution of galaxy color as a function of redshift can be interpreted in terms of simple models for spectral evolution and B/D ratios from the light-profile analysis. However, a few galaxies (<10%) are much redder than the upper envelope of conventional models and are likely to have high internal absorption due to dust lanes and/or dust associated with galaxy mergers. (3) Three galaxies in this sample are close mergers with separations between 0".7 and 1'.0. The occurrence of such systems is much larger than found in previous studies of HST Cycle 1-2 images.