Despite utilization concerns, little information is available on opioid prescribing for acute, disabling low back pain (LBP) and how opioid features (purity, strength, and length of action) and dose change over time. This information is important in targeting guideline implementation efforts and identifying risks for inappropriate prescribing. Using 2002-2003 United States' workers compensation claims, a cohort of 2868 cases with a new episode of work-related LBP and at least one opioid prescription was followed for 2 years. Opioid prescriptions (timing, dose, and formulation), demographics, and medical data were captured. A longitudinal model of change was used to evaluate factors associated with dosing changes. Opioid prescribing typically began early in the course of care (median=8 days, Inter-Quartile Range (IQR)=3, 43 days) and was often prolonged (median=46 days, IQR=14, 329). At the end of the observation period, 7.1% of non-surgical cases and 30.6% of surgical cases were still receiving opioids. The number of days between the initial LBP report and the first opioid prescription had the greatest association with subsequent dose escalation. Dose escalation was greater with pure formulations, and was not related to clinical severity or surgery. In contrast to previous and current guideline recommendations, opioid prescribing for acute LBP was often prolonged, and longer for surgical cases. These results reinforce recommendations to limit opioid duration, and suggest that consideration of opioid features, purity as an important one, can be part of a strategy to prevent escalating dosages.