Religious harassment claims in the United States have risen sharply over the past decade. However, victims of religious harassment may not always report harassment, and true rates may be higher. Hence, actions taken by third parties present (observers) are important in combating harassment in the workplace. The purpose of this paper is to extend a previous model of observer intervention and related research by testing it empirically in the context of religious harassment and identify factors that influence observers' decision to intervene (intervention), when they intervene (level of immediacy), and how much they intervene (level of involvement). Across two studies, we find evidence that verbal harassment, ambiguity of intent, relationship to target/harasser, recurrence belief, religious commitment, pro‐social orientation, and the interactive effect of shared religion and religious commitment predict intervention. Furthermore, individuals show higher levels of involvement and immediacy in intervention when costs are low and emotional reactions are high. Implications of these findings for engaging observers in combatting harassment are discussed. Copyright © 2015 John Wiley & Sons, Ltd.