In this study, we hypothesized that spot urine can be used to predict protein intake at both group and individual levels. Participants (n = 369) of this study were recruited from all 47 prefectures in Japan. Sex-specific formulas were developed based on the ratio of urea nitrogen to creatinine concentration obtained from 3 spot urine samples. Validity of the formulas was examined against two 24-hour urine collections for 7 combinations of spot urine (single and means of 2 or 3 samples) using t test (mean estimation), Spearman correlation, and Bland-Altman plot (individual bias). Means of measured protein intake based on 24-hour urinary excretions were 87.3 g/d (standard deviation 19.7) for men and 70.5 g/d (standard deviation 14.7) for women. Irrespective of sex, the predicted intakes were not significantly different (within 2.7% of differences) from those measured by urinary excretions. Predicted intakes were moderately correlated with measured intakes (men, 0.45-0.60; women, 0.35-0.53). Even after using the mean of 3 samples, Bland-Altman plots showed a considerably wide limit of agreement (men, −30 to 33 g/d; women, −27 to 24 g/d). Except for using single spot urine samples in women, the formula tended to overestimate intake at a lower and underestimate at a higher level of protein intake (slope: men, −0.47 [P < .0001]; women, −0.38 [P = .002]). In conclusion, predictive formulas developed in this study can be used to predict protein intake at group level or to rank individuals' intake but not to predict absolute intake at individual level.