Human immunodeficiency virus (HIV) DNA polymerase chain reaction (PCR) test using venous blood sample has been used for many years in low resource settings for early infant diagnosis of HIV infection in children less than 18 months. The aim of this study was to evaluate and compare the performance characteristics of Amplicor HIV-1 DNA assay version 1.5 following processing of venous blood and dried blood spot (DBS) samples by Roche manual DNA extraction and automated Roche MagNA Pure LC instrument (MP) for HIV-1 DNA PCR testing in Dar es Salaam, Tanzania, in order to scale up early infant diagnosis of HIV infection in routine practice. Venous blood samples from children under 18 months born to HIV-infected mothers between January and April 2008 were collected. Venous blood was used to prepare cell pellet and DBS samples. DNA extractions by manual procedure and MP were performed each on cell pellet, venous blood and DBS samples and tested by Amplicor HIV-1 DNA assay. Of 325 samples included, 60 (18.5%) were confirmed HIV-infected by manual extraction performed on cell pellets. Sensitivity of the assay following MP processing of venous blood was 95% (95% CI; 86.1-99.0%) and 98.3% (95% CI; 91.1 to 99.9%) for the manual extraction and processing by MP performed on DBS samples. Specificity of the assay with all DNA extraction methods was 99.6% (95% CI; 97.9 to 100%). Performance of the assay with Roche manual extraction and processing by MP on DBS samples compared well with Roche manual extraction performed on cell pellet samples. The choice of DNA extraction method needs to be individualized based on the level of laboratory facility, volume of testing and cost benefit analysis before it is adopted for use.