Reporting delay is a common phenomenon in disease case surveillance systems. Accounting for this delay is critical in characterizing the distribution of the disease of interest and, in particular, identifying the recent trend of the disease under surveillance. In this paper, we consider delays in reporting diagnosed HIV cases in the United States. We propose an approach that simplifies and reduces the bias associated with the procedure previously used by CDC for estimating reporting delay probabilities and generating reporting delay weights. The improvements include (1) simplifying the procedure for forming homogeneous reporting delay groups and eliminating the assumption that the reverse-time reporting delay hazards are proportional, (2) simplifying the procedure for estimating the reporting delay probabilities and eliminating the assumptions that no delay exceeds five years and that the reporting delay pattern has not changed in the past five years, and (3) reducing the bias associated with using the inverse of the estimated reporting delay probability as the reporting delay weight. Advantages of the new approach are demonstrated with analytic results and an example using AIDS case surveillance data. These methods can be readily adapted to any surveillance systems where the first or only occurrence of a diagnosis is of interest, covariate information is available to improve estimation of delays in case reporting, and adequate information is available to describe historical patterns in delays in case reporting.