We present a systematic study of proton linewidths in rigid solids as a function of sample spinning frequency and proton density, with the latter controlled by the ratio of protonated and perdeuterated model compounds. We find that the linewidth correlates more closely with the overall proton density (rho(H)) than the size of local clusters of (1)H spins. At relatively high magic-angle spinning (MAS) rates, the linewidth depends linearly upon the inverse MAS rate. In the limit of infinite spinning rate and/or zero proton concentration, the linewidth extrapolates to a non-zero value, owing to contributions from scalar couplings, chemical shift dispersion, and B(0) field inhomogeneity. The slope of this line depends on the overall concentration of unexchangeable protons in the sample and the spinning rate. At up to 30% protonation levels ( approximately 2 (1)H/100A(3)), proton detection experiments are demonstrated to have a substantial (2- to 3-fold) sensitivity gain over corresponding (13)C-detected experiments. Within this range, the absolute sensitivity increases with protonation level; the optimal compromise between sensitivity and resolution is in the range of 20-30% protonation. We illustrate the use of dilute protons for polarization transfer to and from low-gamma spins within 5A, and to be utilized as both magnetization source and detection spins. The intermediate protonation regime enhances relaxation properties, which we expect will enable new types of (1)H correlation pulse sequences to be implemented with improved resolution and sensitivity.