In July of 1987, the year the first HIV Vaccine Clinical Trial Unit was established, a controversial paper showed the effect of zidovudine (AZT) on the frequency of opportunistic infections in HIV-infected men (1). People argued that the effect on viral load was modest and the benefits were not sustained. The report indicated that ‘opportunistic infections developed in 45 subjects receiving placebo, as compared with 24 receiving AZT’. While that statement does not seem earth-shattering, the Kaplan-Meier curves in Figure 1 shouted, ‘treatment of HIV-1 is possible!’ Fortunately, data even when slow in coming, speaks louder than speculation, bias, and dogma. Over the last two decades the determined efforts of thousands of investigators and volunteers have given us more than 25 anti-retroviral drugs, and effective treatment of HIV-1 infection has become routine. There has also been enormous advances in our understanding of pathogenesis and more precision in our characterization of the virus and immune responses to the virus (2). However, vaccine development has been slow and discouraging, punctuated by the report of enhanced infection rates among vaccinees in the Step study (3). While 74 infections in placebo recipients and 51 infections in vaccinees may not sound like a reason for celebration, the Kaplan-Meier curves tell us that ‘vaccine prevention of HIV-1 is possible!’ The report describes a vaccine trial performed in a general population cohort of 16,402 people in Thailand, with a relatively low incidence of infection. The vaccine consisted of a recombinant canarypox vector expressing Envelope, Gag, and parts of Pol and Nef given at 0, 1, 3, and 6 months, rgp120 given at months 3 and 6. The canarypox vector has been evaluated in clinical trial since 1991. It was considered more clinically acceptable than live recombinant vaccinia which had shown promising results in subjects boosted with rgp160 (4) and did not have the problem of pre-existing vector-specific immunity. Multiple Phase I and Phase II trials were performed over the next decade to define the best dose, schedule, and vector design. These studies were done in both the AIDS Vaccine Evaluation Group (AVEG) and the U.S. Military HIV Research Program (USMHRP). The AVEG trials were done in the Western hemisphere and the USMHRP studies were done primarily in Thailand. These studies were carried out during a time of great change in the way T cell function was measured and in our concepts of antibody neutralization. The U.S. studies evaluated a canarypox vector that included not only the HIV antigens, but also E3L and K3L genes from vaccinia (vCP1452). These genes encoded proteins associated with inhibition of apoptosis and in vitro prolonged antigen expression from transduced cells. The importance of apoptosis in antigen presentation and T cell response was not appreciated at the time. In the Phase II study AVEG (HVTN) 203 that evaluated several schedules of vCP1452 + rgp120, immunogenicity did not meet the pre-determined criteria for proceeding to an efficacy trial (5). The criteria for advancing were based on a statistical analysis of whether selected immune responses were frequent enough to define an immune correlate if efficacy was observed in a subsequent efficacy trial. The data indicated that an earlier simpler version of the ALVAC vector (vCP205) was more immunogenic. The studies in Thailand proceeded with a vCP205-style canarypox vector (vCP1521) and continued to meet the predetermined metrics for advancing to efficacy. These studies illustrate the important concept in clinical development that efficacy trials should be designed to answer a specific scientific hypothesis, and should be advanced based on predetermined criteria and a statistical analysis plan that will allow the investigators to answer the primary study hypothesis. Clinical trials cannot answer all questions, especially those asked in retrospect, but when properly designed should accomplish the primary objective of the trial. One of the important lessons from the Thai trial has been learned many times before - human data trumps everything else. Animal models are not interpretable without efficacy data in humans. In retrospect, the low-dose oral challenge studies in infant macaques (6) may have been the closest to anticipating the results of the Thai trial. However, when they were reported these studies were not given serious attention, because the outcome did not meet expectations. It is critical that the scientific community rise above self-serving rhetoric and allow the vaccine development process to be data driven. When the discourse becomes personal and political, scientists lose credibility, and progress is delayed. The history of vaccine development repeatedly demonstrates the value of empirical clinical investigation when the available preclinical data does not provide definitive answers. Both basic and clinical hypothesis-driven research must go forward together. Without parallel testing of hypotheses in human efficacy trials, basic research will remain hypothesis-generating, and will describe the problem of HIV-1 with successively greater precision, but will never reach a conclusion. The real problem is a lack of resources to do all the things that need to be done. There should not have to be a war between different factions of scientific inquiry when human lives are at stake. In the face of public health imperatives, governments need to find new paradigms for supporting both research and development, and to manage the cultural conflicts and competing priorities that will inevitably arise. The greatest point of weakness for HIV is in its transmission efficiency. Despite its success, HIV is not a highly contagious virus, and transmission across mucosal surfaces most often involves a single virion (7). This bottleneck at the point of entry is the Battle of Thermopylae for prevention strategies. Once the virus is replicating in lymph nodes, HIV is highly adapted to evade immune clearance and achieving small reductions in viral load to delay disease progression should be a fallback strategy, and more effectively accomplished with anti-retroviral therapy. Future efforts should be focused on how to reduce transmission efficiency in mucosal tissue with vaccine-induced antibody, T cells, and novel effector mechanisms. The prior VaxGen rgp120 studies in men who have sex with men (MSM) and Thai intravenous drug user (IVDU) populations induced antibody and CD4 T cell responses comparable to those induced in the current study, and did not protect. This raises several questions about trial design and protective immunity that need to be addressed in future studies including risk cohorts recruited for vaccine evaluation, characteristics of the transmitted virus, functional properties of antibody, T cell function and localization, gender and transmission routes. In particular, Does type of risk for infection affect the likelihood of achieving vaccine-induced protection? Circumcision reduced infection rates in African heterosexual transmission cohorts, but has not shown the same type of benefit in MSM cohorts. This is consistent with risk of infection being related to inoculum size and relatively low-dose exposures having a lower threshold of immunity needed for protection. Are there regional differences in the homogeneity and susceptibility of common transmitted strains of HIV? In Thailand, the epidemic is maintained with a relatively homogenous population of viruses based on a common recombinant form combining sequences from subtypes B and E. It is possible that these viruses have different properties than those from other regions, just as challenge stocks of SIVmac251 and SIVE660 appear to have different properties in the macaque model. Is it possible that antibody without classical neutralizing activity can reduce HIV infectivity by slowing mobility in mucosal secretions? This question could potentially be answered by establishing low-dose rectal and vaginal challenge models in macaques using SIV or SHIV and evaluating a dose-range of passively administered antibody with different functional properties. Are there T cell mechanisms that are currently not measured that can result in abortive HIV-1 infection? This would require rapid activation and amplification of effector mechanisms in mucosal tissue in order to clear virus before latency is established. Are there gender differences in the likelihood of vaccine-induced protection? When HIV-1 infects women across vaginal epithelium, the virus has to traverse mucosal secretions of various levels of viscosity that are high in immunoglobulin. Men infected during insertive intercourse do not have that protective barrier. When infection occurs across the rectal mucosa there is more likely to be epithelial disruption, less viscosity in the mucociliary blanket, and a higher density of CD4+ cell targets. Therefore, future efficacy trials must include women at risk of infection through heterosexual transmission. Twenty years from now, I hope we can look back and note this moment as a turning point in the pursuit of an AIDS vaccine. Data from human efficacy trials should promote the development of more authentic and relevant animal models, redirect our scientific objectives, and improve future trial designs. Knowing that vaccine-induced immunity is possible should be an organizing and inspiring principle that leads to stronger collaborative efforts across scientific disciplines and bolsters our political will to develop an effective vaccine.