Preclinical studies have attributed neuroprotective properties to the antibiotic minocycline. Animal studies and early clinical trials support its use in several neurological diseases. In animal spinal cord injury models, minocycline improved neurological and histological outcomes, reduced neuronal and oligodendroglial apoptosis, decreased microglial activation and reduced inflammation. A single-centre, human, double-blind, randomized, placebo-controlled study of minocycline administration after spinal cord injury was undertaken for the purposes of dose optimization, safety assessment and to estimate outcome changes and variance. Neurological, functional, pharmacological and adverse event outcomes were compared between subjects administered 7 days of intravenous minocycline (n = 27) or placebo (n = 25) after acute traumatic spinal cord injury. The secondary outcome used to assess neurological differences between groups that may warrant further investigation was motor recovery over 1 year using the American Spinal Cord Injury Association examination. Recruitment and analyses were stratified by injury severity and injury location a priori given the expected influence of these on the sensitivity of the motor exam. Minocycline administered at higher than previously reported human doses produced steady-state concentrations of 12.7 µg/ml (95% confidence interval 11.6-13.8) in serum and 2.3 µg/ml (95% confidence interval 2.1-2.5) in cerebrospinal fluid, mimicking efficacious serum levels measured in animal studies. Transient elevation of serum liver enzymes in one patient was the only adverse event likely related to the study drug. Overall, patients treated with minocycline experienced six points greater motor recovery than those receiving placebo (95% confidence interval -3 to 14; P = 0.20, n = 44). No difference in recovery was observed for thoracic spinal cord injury (n = 16). A difference of 14 motor points that approached significance was observed in patients with cervical injury (95% confidence interval 0-28; P = 0.05, n = 25). Patients with cervical motor-incomplete injury may have experienced a larger difference (results not statistically significant, n = 9). Functional outcomes exhibited differences that lacked statistical significance but that may be suggestive of improvement in patients receiving the study drug. The minocycline regimen established in this study proved feasible, safe and was associated with a tendency towards improvement across several outcome measures. Although this study does not establish the efficacy of minocycline in spinal cord injury the findings are encouraging and warrant further investigation in a multi-centre phase III trial. ClinicalTrials.gov number NCT00559494.