How do news media cover civilian suffering in foreign wars and how does the public react to it? We study the full pipeline from editorial choices to public attitudes in the context of the Gaza war. We analyze a corpus of roughly 150,000 news articles from 30 English-language outlets over two years of the conflict. We introduce a continuous measure of civilian-centeredness of news coverage, built from crowdsourced ratings and a fine-tuned language model. Supply is starkly unequal: each Israeli civilian death generates roughly 39 times as many headlines as each Palestinian civilian death. Demand presents a puzzle: we document using web-browsing data  that Americans read less civilian-centered coverage relative to the supply available. Yet, in a randomized headline-choice experiment they prefer coverage of civilians. We reconcile these divergent findings in a structural model of news consumption with unobserved choice sets. The model estimates suggest that a one-standard-deviation increase in civilian-centeredness reduces an article’s chance of reaching readers by roughly a fifth; this offsets the public’s taste for such coverage. Americans consume less civilian-centered war coverage than they want, not because they look away from suffering, but because that coverage is less likely to reach them. A pilot survey experiment suggests exposure to civilian-centered coverage moves attitudes in a humanizing direction, though effects are imprecisely estimated.