We use Natural Language Processing (NLP) techniques to examine if judgments’ complexity in Supreme Court cases influences dowry deaths in India. Leveraging the quasi-random assignment of judges to cases, we find that a one-unit increase in the complexity of judgments captured by the Fog Index in a state year increases dowry deaths by 2% in the subsequent period. Based on estimates of the Value of Statistical Life (VSL), this increase in dowry-related homicides of women costs $1.83 million in a state year or 51 million USD for the country annually. In otherwise similar judgments, very high complexity increases the incidence of future dowry deaths by 15%. However, having women justices on the panel of judges in such cases mitigates the effect of higher complexity.