The main protease (Mpro) is a validated antiviral drug target of SARS-CoV-2. A number of Mpro inhibitors have now advanced to animal model study and human clinical trials. However, one issue yet to be addressed is the target selectivity over host proteases such as cathepsin L. In this study we describe the rational design of covalent SARS-CoV-2 Mpro inhibitors with novel cysteine reactive warheads including dichloroacetamide, dibromoacetamide, tribromoacetamide, 2-bromo-2,2-dichloroacetamide, and 2-chloro-2,2-dibromoacetamide. The promising lead candidates Jun9-62-2R (dichloroacetamide) and Jun9-88-6R (tribromoacetamide) had not only potent enzymatic inhibition and antiviral activity but also significantly improved target specificity over caplain and cathepsins. Compared to GC-376, these new compounds did not inhibit the host cysteine proteases including calpain I, cathepsin B, cathepsin K, cathepsin L, and caspase-3. To the best of our knowledge, they are among the most selective covalent Mpro inhibitors reported thus far. The cocrystal structures of SARS-CoV-2 Mpro with Jun9-62-2R and Jun9-57-3R reaffirmed our design hypothesis, showing that both compounds form a covalent adduct with the catalytic C145. Overall, these novel compounds represent valuable chemical probes for target validation and drug candidates for further development as SARS-CoV-2 antivirals.