This paper describes several models to convert order-of-magnitude count data that are collected during REEF Roving Diver Technique (RDT) surveys to a numeric mean, and demonstrate that with a sufficient number of surveys, estimates of the mean with a reasonably small confidence interval can be attained. For each model, parameter estimates and associated confidence intervals were derived from 292 RDT surveys where precise counts were also made. Models were compared using the small sample Akaike Information Criteria (AICc). The best-fitting model uses disaggregated bin-count data and considers the relative proportion of counts in adjacent bins. A companion paper was published in the same CalCOFI issue that uses the model to evaluate population trends in rocky reef fish species along the Monterey Peninsula region in central California.