Objective: To test the efficacy of low-level laser therapy on lateral ankle sprains as an addition to a standardized treatment regimen, a trial was conducted in which high-dose laser (5J/cm2), low-dose laser (0,5J/cm2), and placebo laser therapy (0J/cm2) at skin level were compared. Design: Randomized, double-blind, controlled clinical trial with a follow-up of 1 year. Patients, therapists, assessors, and analysts were blinded to the assigned treatment. Setting: An ambulatory care setting. Patients: After informed consent and verification of exclusion criteria, 217 patients with acute lateral ankle sprains were randomized to three groups from September 1, 1993, through December 31, 1995. Interventions: Twelve treatments of 904nm laser therapy in 4 weeks as an adjunct to a standardized treatment regimen of 4 weeks of brace therapy combined with standardized home exercises and advice. The laser therapy device used was a 904nm Ga-As laser, with 25-watt peak power and 5,000 or 500Hz frequency, a pulse duration of 200nsec, and an irradiated area of 1cm2. Primary Outcome Measures: Pain and function as reported by the patient. Results: Intention-to-treat analysis of the short- term results showed no statistically significant difference on the primary outcome measure, pain (p = .41), although the placebo group showed slightly less pain. Function was significantly better in the placebo group at 10 days (p = .01) and 14 days (p = .03) after randomization. The placebo group also performed significantly better on days of sick leave (p = .02) and at some points for hindrance in activities in daily life and pressure pain, as well as subjective recovery (p = .05). Intention-to-treat analysis showed that total days of absenteeism from work and sports were remarkably lower in the placebo group than in the laser groups, ranging from 3.7 to 5.3 and 6 to 8 days, respectively. The total number of relapses at 1 year in the low-dose laser group (n = 22) was significantly higher (p = .04) than in the other two groups (high laser, n = 13; placebo, n = 13). Subgroup analysis to correct for possible confounders did not alter these findings.