Differential unit non‐response in household wealth surveys biases estimates of top tail wealth shares downward. Using Monte Carlo evidence, I show that adding only a few extreme observations to wealth surveys is sufficient to remove the downward bias. Combining extreme wealth observations from Forbes World's billionaires with the Survey of Consumer Finances, the Wealth and Assets Survey, and the Household Finance and Consumption Survey, I provide new improved estimates of top tail wealth in the United States, the United Kingdom, and nine euro area countries. These new estimates indicate significantly higher top wealth shares than those calculated from the wealth surveys alone.