The focus of this paper is on the development of methodology for short-term (1–3 days) oceanic bioluminescence (BL) predictions and the optimization of spatial and temporal bioluminescence sampling strategies. The approach is based on predictions of bioluminescence with an advection–diffusion–reaction (tracer) model with velocities and diffusivities from a circulation model. In previous research, it was shown that short-term changes in some of the salient features in coastal bioluminescence can be explained and predicted by using this approach. At the same time, it was demonstrated that optimization of bioluminescence sampling prior to the forecast is critical for successful short-term BL predictions with the tracer model. In the present paper, the adjoint to the tracer model is used to study the sensitivity of the modeled bioluminescence distributions to the sampling strategies for BL. The locations and times of bioluminescence sampling prior to the forecast are determined by using the adjoint-based sensitivity maps. The approach is tested with bioluminescence observations collected during August 2000 and 2003 in the Monterey Bay, California, area. During August 2000, BL surveys were collected during a strong wind relaxation event, while in August 2003, BL surveys were conducted during an extended (longer than a week) upwelling-favorable event. The numerical bioluminescence predictability experiments demonstrated a close agreement between observed and model-predicted short-term spatial and temporal changes of the coastal bioluminescence.