The paper presents a specialised Partial Least Squares (PLS) algorithm for on-line calibration monitoring. At first, the basic scheme and philosophy of PEANO (Process Evaluation and Analysis by Neural Operators), an on-line calibration monitoring system, is briefly discussed. Then the reader gets introduced into the general idea of PLS, followed by an extension to Neural Network PLS (NNPLS). The main part of this paper will present specification and finding from using an enhanced NNPLS method. The new type of algorithm, referred to as NNPLS-t, can deal with time-delayed signals and has several advantages in comparison to the standard NNPLS method. Results of both techniques are presented and compared.