Background: the advantages of using simulators in skills training are generally recognized, but simulators are often too expensive for medical schools in developing countries. Cheaper locally-made models (or part-task trainers) could be the answer, especially when teachers are involved in design and production (teacher-made models, tm). Methods. We evaluated the effectiveness of a tm in training and assessing intravenous injection skills in comparison to an available commercial model (cm) in a randomized, blind, pretest-posttest study with 144 undergraduate nursing students. All students were assessed on both the tm and the cm in the pre-test and post-test. After the post-test the students were also assessed while performing the skill on real patients. Results: differences in the mean scores pre- and post-test were marked in all groups. Training with tm or cm improved student scores substantially but there was no significant difference in mean scores whether students had practiced on tm or cm. Students who practiced on tm performed better on communication with the patient than did students who practiced on cm. Decreasing the ratio of students per tm model helped to increase practice opportunities but did not improve student’s mean scores. The result of the assessment on both the tm and the cm had a low correlation with the results of the assessment on real persons. Conclusions: the tm appears to be an effective alternative to cm for training students on basic iv skills, as students showed similar increases in performance scores after training on models that cost considerably less than commercially available models. These models could be produced using locally available materials in most countries, including those with limited resources to invest in medical education and skills laboratories. 2012 Tran et al.; Licensee biomed central ltd.