This study compared five density estimation techniques applied to samples from a population of 272,244 examinees' ACT English Usage and Mathematics Usage raw scores. Unsmoothed frequencies, kernel method, negative hypergeometric, four-parameter beta compound binomial, and Cureton-Tukey methods were applied to 500 replications of random samples of 500, 1000, 2000, and 5000 from these populations. The four-parameter beta compound binomial produced the most accurate estimates, and the kernel method yielded only slightly less accurate estimates. Cureton-Tukey ranked third in accuracy. All methods involving smoothing produced more accurate estimates than unsmoothed frequencies except the negative hypergeometric. Negative hypergeometric estimates varied erratically by test and score level. The methods studied have the potential to improve the estimation of norms and the equipercentile equating function. (Author)