Predation by feral cats (Felis catus) is believed to threaten the conservation of a range of terrestrial vertebrates in southeastern Australia, and new baiting techniques are sought for broad-scale control of feral cats. In southeastern Australia there are 34 native mammals that may consume meat baits, and ways to minimize their exposure to bait toxicants are needed. We determined the potential of cats to ingest larger particles relative to most nontarget species as a mechanism to increase baiting selectivity. Feral cats reliably ingested inert, spherical bearings up to 4.7 mm in diameter when the bearings were implanted within a specialized bait medium. Presence of bearings did not affect bait consumption relative to untreated baits. Repetitive ingestion was highly reliable in the first 9 days of a feeding trial and diminished only marginally in a consecutive trial. We presented captive plains rats (Pseudomys australis), fat-tailed dunnarts (Sminthopsis crassicaudata), eastern barred-bandicoots (Perameles gunnii), and northern quolls (Dasyurus hallucatus), with baits containing 4.7-mm-diameter coated pellets formulated with the marker dye rhodamine B (RB). Exposure to the marker dye for each species was not biased to individual or day of presentation. Exposure to RB in the pellet occurred in only 3.1–6.5% of presentations for each species, and the mean daily mass of the pellet in g kg⁻¹ day⁻¹ ingested was 0.078–0.01% of the mean bait mass in g kg⁻¹ day⁻¹ consumed. Pellet presentation greatly reduced (P ≤ 0.001) the exposure of wild native rodents to RB relative to directly injected baits. Exploiting differential particle size ingestion between feral cats and nontarget species could potentially reduce exposure of many nontarget mammals to bait toxicants and decrease the risk of baiting to nontarget species.