We observed no association between the distribution of fitness of members with DENV-1 populations in mosquito cells and the survival or extinction of a lineage of viruses. However, there was an increase in the proportion of more fit members during and after the 2001-2002 outbreaks, suggesting that there may have been some selection for DENV that grew to high titre in mosquitoes. This observation has two caveats. The first is that there was an increase in the proportion of more fit members in population 36957/00, compared to 31459/98 before the outbreak began. The second is that the half life of an Aedes aegypti mosquito in Thailand is only 7–8 days and so selection might be for a virus that replicated faster rather than one that grew to high titre. However, DENV, which grows to high titre, may also reach significant titres earlier. Fitness can be defined as a measure of the ability to replicate in a host but a more appropriate definition could be a measure of the ability to be transmitted, i.e. to infect the next host in a transmission cycle. While transmissibility is probably the most relevant measure for a virus like DENV, with infection cycles involving alternate human and mosquito hosts, technical constraints prevented this measure being used. In this study, fitness was defined as the yield of DENV1 virions from infected cells. An indirect ELISA procedure was employed to estimate the quantity of DENV virions released into a culture supernatant. It was accepted that a proportion of these would contain TH-302 genomes that were not infectious. However, given the complexity of the interactions between RNA genomes e.g. complementation, interference by sub-genomic RNA etc., the yield of virions was a more relevant measure of productive infection than estimates of either the number of infectious virus particles or of genome copy number. A comparison of titres of DENV in patients measured as infectious virus or as genome copy number suggested that copy number values are 10–100 times higher than infectious titres. That the individual members of DENV populations might vary in their fitness is not surprising given that there are reports of virions with genomes with mutations and indels giving rise to intragenic stop codons as well as genomes with deletions of thousands of nucleotides There also is an extensive literature describing non-lethal changes that effect DENV replication. Taken together with the comments above, it is unlikely that an individual cell is infected by a single DENV genome. For these reasons, a unit, “one infectious dose” has been used in this study and has been derived statistically, i.e. if 96 infectious units of virus in 9.6 ml are aliquoted uniformly into 96 wells of a microtitre plate, only 65 wells will contain virus. While this is a weakness of this approach, there was no alternative, and the same methodology was used for all populations, so enabling comparisons to be made.