No empirical study on court-adjudicated condemnation compensation has been done in the past 30 years. To fill in the empirical gap, I first collect condemnation compensation cases in New York City between 1990 and 2003, finding that the court usually rules in favor of condemnors. Additionally, I use hedonic regression models and about 7,500 sales to estimate the fair market value (FMV) of 27 condemned properties and compare the FMV with condemnors' offered value, condemnees' claimed value, and court awards, finding that all condemnees' claims are above FMV, while many condemnors' offers and court awards are above FMV; moreover, condemnors' claims come closer to FMV than condemnees' claims and court awards. The court usually favors condemnors because condemnors have better incentives to produce unbiased assessed value and condemnors' offered value actually is closer to FMV. Offers, claims, and awards are inaccurate because bias-prone appraisal techniques have been used to assess property value.