Oral methotrexate is the benchmark against which other disease-modifying anti rheumatic drugs are measured. The use of parenteral methotrexate for those failing to tolerate or respond to oral therapy is accepted, but indications for its use and its place in the therapeutic ladder have not been fully investigated. We assessed the use of parenteral methotrexate (MTX) in our rheumatoid arthritis (RA) population and compared the characteristics of these patients to a matched group of those on oral therapy. We compared response rates to each approach using DAS 28 scores, ESR and visual analogue scales. Inferences on costs of parenteral therapy were made and predictors of response defined. We found that 10% of our total RA patient population were on parenteral methotrexate, having failed to tolerate or respond to oral therapy. Seventy-five percent of these met the criteria for the use of anti-tumour necrosis factor (TNF) agents. Overall response rates were equivalent to those obtained by responders to oral MTX. Patients on parenteral therapy were younger and were more likely to have extreme values of body mass index (BMI) than those on oral therapy. The approach was economically viable, although many patients unnecessarily attended hospital to receive their injections. We advocate consideration of parenteral MTX in all RA patients unresponsive to oral therapy prior to treatment with anti-TNF therapy. Response to parenteral therapy can be predicted by low BMI (below 22 kg/m(2)), possibly as a result of malabsorption, or by high BMI (over 30) as a result of gastrointestinal intolerance. A mechanism to deliver this option through self-administration in the community should be encouraged.