Molecular property prediction is a critical task in drug discovery, aiding in the differentiation between promising and ineffective drug-like compounds. While machine learning (ML) methods have been employed to predict the biological activity of drug candidates, capturing more complex and highly discriminative features from molecular data has led to the adoption of advanced deep learning (DL) techniques. In this work, we propose a deep convolutional Spiking Neural Network (SNN) architecture, fpCSNN, that utilizes molecular fingerprints to encode the chemical structure of compounds as binary vectors for molecular property prediction. By converting molecular fingerprints into spike trains, SNNs are able to process the molecular data in a temporal encoded format. This approach enables the extraction of local dependencies from neural spike signals to predict molecular properties, such as chemical toxicity and potential drug side effects. We validate fpCSNN’s accuracy and robustness through experiments on benchmark datasets, including Tox21 and SIDER, outperforming standard ML baselines. Future work explores additional spiking models, hyperparameter optimization, and more datasets. Data and code are available at: https://github.com/Dinu-Bosii/fpCSNN .