This paper presents the development of an intelligent temperature to frequency converter to measure the temperature using a negative temperature coefficient (NTC) thermistor. The signal conditioning circuit (SCC) of the NTC thermistor is a modified timer circuit whose control voltage is generated by a difference amplifier. The thermistor SCC acts as a temperature to frequency converter and exhibits a moderate linear temperature-frequency characteristic over a range of 0-100 °C with a linearity error of ±3.5%. A multilayer perceptron (MLP) neural network with Levenberg-Marquardt (LM) algorithm is used for modeling and nonlinearity estimation of the converter. The LM algorithm effectively reduces the linearity error as compared to the Broyden-Fletcher-Goldfarb-Shanno (BFGS) and the scaled conjugate gradient (SCG) algorithms. Mean square error (MSE), regression coefficients, linearity, accuracy and dispersion spread are used to evaluate the performance of the proposed ANN-based modeling. The intelligence of the ANN-based modeling is embedded in a low cost microcontroller unit and the performance is experimentally verified on a prototype unit. The linearity error and sensitivity of the proposed unit are approximately ±0.35% and 5 kHz/°C respectively. © 2016 Elsevier Ltd. All rights reserved.