BACKGROUND: About 13 years since the introduction of antiretroviral therapy (ART) for children living with HIV (CLHIV) in Zambia, HIV/AIDS testing and treatment guidelines for children have evolved over the years with limited documentation of long-term trends in the numbers testing HIV positive and initiating ART. We examined trends in HIV testing and ART initiation in Zambia. METHODS: We conducted a retrospective cohort study using routinely collected patient level data from 496 health facilities across Zambia. We used Poisson regression to derive incident rate ratios and 95% confidence intervals (95% CI) for background characteristics and used a Cuzick non-parametric test for trends to test the 13-year trends. Median time from testing to ART initiation in days and incidence rates were derived using life tables in survival analysis. We used multi-level random effects Poisson regression model to determine variations in time from HIV testing to ART initiation by facility. RESULTS: Overall, the cumulative proportion of the children who tested positive and initiated antiretroviral therapy (ART for HIV) from 2004 to 2017 was 69% (n = 99 592). During the period under review proportions of ART initiation increased from 52% in 2004-2006 to 97% in 2016-2017 (P < 0.001) and time from testing to ART initiation reduced from a median of 17 days IQR (1-161) in 2004 to one day IQR (1-14), P < 0.001 in 2016-2017. CLHIV were 15 times more likely to be initiated on ART in 2016-17 compared to period 2004-6 (IRR = 15.2, 95% CI 14.7-15.7). Time to ART initiation increased with age and was higher in rural health facilities compared to urban facilities. About 11% of the variability in time to ART initiation in children could be attributed to differences between facilities. CONCLUSIONS: The substantial increase in ART initiation and reduction in time to ART initiation among CLHIV identified in this study, reflects improvements in the paediatric HIV programme in Zambia in relation to health care delivery and adherence to national testing and treatment guidelines that were adapted from WHO guidelines. However, age-related differentials in rates of ART initiation suggests that urgent interventions are needed to sustain and further improve programme performance.