Small laboratory cage trials of non-drive and gene-drive strains of the Asian malaria vector mosquito, Anopheles stephensi, were used to investigate release ratios and other strain properties for their impact on transgene spread during simulated population modification. We evaluated the effects of transgenes on survival, male contributions to next-generation populations, female reproductive success and the impact of accumulation of gene drive-resistant genomic target sites resulting from nonhomologous end-joining (NHEJ) mutagenesis during Cas9, guide RNA-mediated cleavage. Experiments with a non-drive, autosomally-linked malaria-resistance gene cassette showed 'full introduction' (100% of the insects have at least one copy of the transgene) within 8 weeks (≤ 3 generations) following weekly releases of 10:1 transgenic:wild-type males in an overlapping generation trial design. Male release ratios of 1:1 resulted in cages where mosquitoes with at least one copy of the transgene fluctuated around 50%. In comparison, two of three cages in which the malaria-resistance genes were linked to a gene-drive system in an overlapping generation, single 1:1 release reached full introduction in 6-8 generations with a third cage at ~80% within the same time. Release ratios of 0.1:1 failed to establish the transgenes. A non-overlapping generation, single-release trial of the same gene-drive strain resulted in two of three cages reaching 100% introduction within 6-12 generations following a 1:1 transgenic:wild-type male release. Two of three cages with 0.33:1 transgenic:wild-type male single releases achieved full introduction in 13-16 generations. All populations exhibiting full introduction went extinct within three generations due to a significant load on females having disruptions of both copies of the target gene, kynurenine hydroxylase. While repeated releases of high-ratio (10:1) non-drive constructs could achieve full introduction, results from the 1:1 release ratios across all experimental designs favor the use of gene drive, both for efficiency and anticipated cost of the control programs.