Selectivity, the ability to kill weeds without killing the crop, is a challenge for in-row mechanical cultivation, especially in slow-growing crops like carrots [Daucus carota L. ssp. sativus Hoffm. ‘Bolero’]. To gain insight into the optimal tool type and timing for in-row cultivation of different weed species, we adapted an existing model (“Kurstjens model”) to predict “potential efficacy” (PE)—the greatest weed mortality attainable at a given level of crop mortality—based on weed anchorage force and height data, which serve as proxies for tolerance to uprooting and burial. We parametrized the baseline model using data for carrots and five weed species at early growth stages and used the model to predict the PE of idealized tools that bury or uproot in combination with various cultural practices. Under baseline model assumptions, tools that bury had greater PE for grass weeds, and tools that uproot had greater PE for broadleaves. Combining or “stacking” tools that uproot with those that bury had minimal impact on predicted PE for individual weed species, but increased PE on mixed grass–broadleaf weed communities compared with single-tool mechanisms of action. Cultural practices (e.g., stale seedbedding and cultivar choice) that increased carrot anchorage force and height relative to weeds at the time of cultivation greatly increased PE for both mechanisms of action. Our model provides a useful method for predicting the optimal tool mechanism of action and timing for any weed–crop combination.