FMRI studies of the orbitofrontal cortex or the inferior temporal lobes are often compromised by susceptibility artefacts, which may result in signal reduction or loss in gradient echo (GE) EPI. Spin echo (SE) EPI is considerably more robust against susceptibility-related signal loss, but its intrinsic sensitivity to changes in the blood oxygenation level dependent (BOLD) contrast is generally lower. In this study, we performed a direct comparison of GE and SE fMRI using a single-shot dual echo EPI acquisition scheme. Transient hypercapnia, induced by breathing Carbogen (5% CO2, 95% O2), was used as a global physiological stimulus to alter the BOLD contrast. In regions affected by magnetic field inhomogeneities, SE EPI provided significantly higher BOLD sensitivity than GE EPI. Such regions included the orbitofrontal cortex, temporal pole, anterior inferior temporal cortex, as well as parts of the lateral inferior temporal cortex and the lateral cerebellum. Dual echo fMRI benefits from the robustness of SE EPI in these critical regions while utilising the generally higher sensitivity of GE EPI in normal regions. It therefore provides an attractive solution for fMRI studies that require optimum sensitivity in both normal and critical brain regions. Furthermore, a general method is proposed to combine the GE and SE data into a single hybrid data set that provides optimum sensitivity in the whole brain. This method can be applied to any experimental design that can be expressed in terms of a generalised linear model. The feasibility of this approach is demonstrated both theoretically and experimentally.