Despite several attempts to develop an effective prophylactic vaccine for HSV-2, all have failed to show efficacy in the clinic. The most recent of these failures was the GlaxoSmithKline (GSK) subunit vaccine based on the glycoprotein gD with the adjuvant monophosphoryl lipid A (MPL). In a phase 3 clinical trial, this vaccine failed to protect from HSV-2 disease, even though good neutralizing antibody responses were elicited. We aimed to develop a superior, novel HSV-2 vaccine containing either gD or gB alone or in combination, together with the potent adjuvant CpG oligodeoxynucleotides (CPG). The immunogenic properties of these vaccines were compared in mice. We show that gB/CPG/alum elicited a neutralizing antibody response similar to that elicited by gD/CPG/alum vaccine but a significantly greater gamma interferon (IFN-γ) T cell response. Furthermore, the combined gB-gD/CPG/alum vaccine elicited significantly greater neutralizing antibody and T cell responses than gD/MPL/alum. The efficacies of these candidate vaccines were compared in the mouse and guinea pig disease models, including a novel male guinea pig genital disease model. These studies demonstrated that increased immune response did not correlate to improved protection. First, despite a lower IFN-γ T cell response, the gD/CPG/alum vaccine was more effective than gB/CPG/alum in mice. Furthermore, the gB-gD/CPG/alum vaccine was no more effective than gD/MPL/alum in mice or male guinea pigs. We conclude that difficulties in correlating immune responses to efficacy in animal models will act as a deterrent to researchers attempting to develop effective HSV vaccines.