Background: Accurate and selective assessment of testosterone requires use of a sensitive LC-MS/MS method, especially at low levels as those seen in young children. Methods: The present longitudinal study of 20 healthy children from the Copenhagen Puberty Study followed every 6 months for 5 years evaluates the longitudinal increase in serum testosterone before, during and after pubertal onset quantified by a newly developed LC-MS/MS method in comparison with immunoassay. Testosterone concentrations in serum samples (n = 177) were determined by LC-MS/MS (detection limit 0.1 nmol/l) and by immunoassay (detection limit 0.23 nmol/l). Results: Serum concentrations of testosterone increased gradually with age by both methods. However, serum testosterone was quantifiable in 9/10 girls prior to pubic hair development measured with LC-MS/MS, and in 2/10 girls measured with immunoassay. In boys, testosterone was quantifiable in 10/10 boys 1 year prior to pubic hair development measured with LC-MS/MS, and only in 1/10 boys measured with immunoassay. Serum testosterone levels were quantifiable 1.5 years (range 0.5-2.5) earlier using LC-MS/MS. Conclusion: Assessment of longitudinal circulating levels of serum testosterone using a selective LC-MS/MS method proved to be more sensitive in predicting early peripubertal changes in healthy children compared to levels determined by immunoassay.