The invention discloses a similarity detection method and a similarity detection device. The similarity detection method comprises the following steps of: acquiring two groups of detection data, wherein the detection data is used for characterizing the product property and accords with logarithmic normal distribution; establishing a linear regression model corresponding to a logarithmic normal distribution function of each group of detection data; calculating the error square and the error mean square corresponding to the linear regression model of each group of detection data; setting a confidence belt corresponding to the linear regression model of each group of detection data, wherein the confidence belt is associated with the calculated error mean square; and determining that the two groups of detection data are similar if the linear regression model of each group of detection data is in a corresponding confidence belt of the linear regression model of the other group of detectiondata. The similarity detection method and the similarity detection device can improve the similarity detection accuracy.