Catalyst poisoning is the partial or total deactivation of a catalyst by a chemical compound. Poisoning refers specifically to chemical deactivation, rather than other mechanisms of catalyst degradation such as thermal decomposition or physical damage. Although usually undesirable, poisoning may be helpful when it results in improved catalyst selectivity (e.g. Lindlar's catalyst). An important historic example was the poisoning of catalytic converters by leaded fuel.

Organic functional groups and inorganic anions often have the ability to strongly adsorb to metal surfaces. Common catalyst poisons include carbon monoxide, halides, cyanides, sulfides, sulfites, phosphates, phosphites and organic molecules such as nitriles, nitro compounds, oximes, and nitrogen-containing heterocycles. Agents vary their catalytic properties because of the nature of the transition metal. Lindlar catalysts are prepared by the reduction of palladium chloride in a slurry of calcium carbonate (CaCO₃) followed by poisoning with lead acetate. In a related case, the Rosenmund reduction of acyl halides to aldehydes, the palladium catalyst (over barium sulfate or calcium carbonate) is intentionally poisoned by the addition of sulfur or quinoline in order to lower the catalyst activity and thereby prevent over-reduction of the aldehyde product to the primary alcohol.

Poisoning often involves compounds that chemically bond to a catalyst's active sites. Poisoning decreases the number of active sites, and the average distance that a reactant molecule must diffuse through the pore structure before undergoing reaction increases as a result. As a result, poisoned sites can no longer alter the rate of reaction. Large scale production of substances such as ammonia in the Haber–Bosch process include steps to remove potential poisons from the product stream. When the poisoning reaction rate is slow relative to the rate of diffusion, the poison will be evenly distributed throughout the catalyst and will result in homogeneous poisoning of the catalyst. Conversely, if the reaction rate is fast compared to the rate of diffusion, a poisoned shell will form on the exterior layers of the catalyst, a situation known as "pore-mouth" poisoning, and the rate of catalytic reaction may become limited by the rate of diffusion through the inactive shell. Homogenous and "pore-mouth" poisoning occurrences are most frequently observed when using a porous medium catalyst.

If the catalyst and reaction conditions are indicative of low effectiveness, selective poisoning may be observed, where poisoning of only a small fraction of the catalyst's surface gives a disproportionately large drop in activity.

If η is the effectiveness factor of the poisoned surface and h_p is the Thiele modulus for the poisoned case:

When the ratio of the reaction rates of the poisoned pore to the unpoisoned pore is considered:

where F is the ratio of poisoned to unpoisoned pores, h_T is the Thiele modulus for the unpoisoned case, and α is the fraction of the surface that is poisoned.

The above equation simplifies depending on the value of h_T. When the surface is available, h_T is negligible: