Effective maritime search and rescue (SAR) requires accurate assessment of vessel distress based on ambiguous and probabilistic signals. This study examined how individuals aggregate positive and null diagnostic signals when making SAR launch decisions. Participants (N = 476) were randomly assigned to one of nine conditions in a 3 (Error Penalty: False Negative [FN] = -100/False Positive [FP] = -100; FN = -150/FP = -50; FN = -50/FP = -150) × 3 (Base Rate: 25%, 50%, or 75% probability of vessel distress) between-subjects design. Each participant completed four scenarios, with the number of distress and benign scenarios determined by their assigned base rate condition: participants in the 25% condition received one distress and three benign scenarios; those in the 50% condition received two of each; and those in the 75% condition received three distress and one benign scenario. Distress scenarios contained 80% positive signals, while benign scenarios contained 80% null signals. Participants could request up to eight signals per scenario before committing to a launch or stand-down decision. Despite this, participants requested an average of only two signals, and decision accuracy did not exceed chance levels in the 25% and 75% base rate conditions. Mixed-effects logistic regression indicated that higher subjective likelihood ratings strongly predicted launch decisions (OR = 7.77), whereas greater expressed confidence was paradoxically associated with reduced launch probability (OR = 0.64). Asymmetric scoring rules that more severely penalized false negatives were associated with increased launching behavior. These findings implicate normalcy bias and base rate neglect as joint contributors to suboptimal SAR decision-making, with implications for operator training and the design of decision-support systems.