A new model named as nonlinear partial least squares (PLS) with slice transformation (SLT) based piecewise linear inner relation (NPLSSLT) is proposed to handle nonlinear calibration. It is based on a nonlinear inner relation combined with a modified error-based weight updating (EBWU) strategy. In NPLSSLT, the SLT based piecewise linear mapping function is used to construct the nonlinear inner relation between the input and output score vectors, and a linear PLS is used to improve the EBWU procedure. This enables the construction of a nonlinear nested PLS model that seems to be an attractive alternative to other nonlinear PLS models. The performance of NPLSSLT is evaluated on two benchmark near-infrared (NIR) datasets and compared with five other models: PLS, Quadratic PLS (QPLS), spline PLS (SPLS), neural network PLS (NNPLS) and SLT based PLS (SLT-PLS). Additionally the Wilcoxon signed rank test was used to statistically compare predictive performance of two competing calibration models. Experimental results showed that NPLSSLT did quite well on dataset 1 and not badly on dataset 2.