Hearing protection devices (HPDs) play a significant role in protecting workers from occupational noise-induced hearing loss. Individual HPD fit-testing estimates the amount of protection, or attenuation, that an individual achieves from a given HPD as it is worn. Results from a single fit-test may not be representative of real-world HPD performance over time, however, due to inconsistency in how the individual fits the HPD from time to time. In this study, the effects of HPD type and user training on the consistency of attenuation achieved across multiple fittings were evaluated in a within-subjects design. Attenuation measurements using a real-ear attenuation at threshold procedure were obtained on 30 participants wearing custom-molded and non-custom earplugs. The subjects were initially naive to proper earplug insertion techniques and later received one-on-one training for the second half of the attenuation measurements. Consistency, or reliability, of fit was assessed using (i) the standard deviation of the 'distance to ear mean attenuation', a measure of fitting uncertainty, and (ii) the standard deviation of the attenuation values across multiple fit-tests for each subject. The custom earplug provided statistically significantly better consistency of attenuation than the non-custom earplug at 500, 1000, and 2000 Hz. Training effects were statistically significant at 250, 500, and 1000 Hz and at the Personal Attenuation Rating. No interactions were statistically significant. These results indicate that, in general, subjects obtained more consistent attenuation with the custom earplugs than with the non-custom earplugs and that consistency improved with training for both earplug types.