The measurement process of attachment loss has been criticized in recent years. Problems with clinical interpretation, precision of the measurement, and statistical manipulation of the obtained data, are some of the problems associated with the present methodology. The purpose of the present study was to propose an alternative measurement process which addresses some of the existing problems by estimating the lost attachment surface area (LAS) and the remaining attachment surface area (RAS) from a combination of clinical measurements. The results show that a linear combination of several sources of clinical information can be used to predict RAS and LAS. A diagnostic model for LAS (R2 = 81.5%) predicts the square root of LAS with information obtained from bucco-lingual attachment level measurements, the radiographic lost attachment area, the gingivitis index and the radiographic tooth length. This model increases the precision of the estimate of LAS by a factor of 1.86 when compared to the estimate of LAS using only attachment level measurements. A diagnostic model for RAS (R2 = 75.5%) predicts the square root of RAS with the information obtained from the remaining radiographic attachment area, the gingivitis index and the mobility index. Both linear inference models are constructed with measurements of anatomical landmarks to avoid the discrepancy between anatomical and clinical measurements in the produced estimates. It is concluded that modeling of periodontal data provides a simple, inexpensive, and precise diagnostic tool for predicting the lost and the remaining periodontal attachment of single-rooted teeth. Measurement processes of this type could provide a convincing basis for the evaluation of clinical decisions and research questions.