In this work the development of affinity sensors for the detection of microcystin-LR based on a computationally designed artificial receptor is presented. Microcystin-LR is a cyclic heptapeptide hepatotoxin produced by Cyanobacteria (aquatic organisms also known as blue-green algae), which during blooms period can release toxins in water. Clinical signs of hepatotoxicosis have been observed in domestic animals and livestock and recently also in humans. At present, analysis of these toxins is achieved largely using conventional, time consuming and expensive techniques such as chromatographic methods (HPLC, TLC) and immunoassay. Therefore, the necessity of an easy and inexpensive method of analysis such a biosensor is becoming urgent. In this work an artificial receptor for microcystin-LR was synthesised using a combined approach of molecular imprinting and computer modelling. A computer-aided rational design was applied to study microcystin-LRlmonomers interactions in order to find an optimal composition for the synthesis of the receptor. The optimised composition, suggested by computer modelling, consisted in 1 mol of2-acrylamido-2-methyl-propanesulfonic acid and 6 mol ofurocanic acid ethyl ester for 1 mol of template. This monomer composition was then used to synthesise a molecularly imprinted polymer (MIP) and an enzyme- linked competitive assay was developed to characterise the computational receptor. In the assay, computational MIP was able both to detect 0.1 ~g rl of microcystin-LR and to distinguish the analyte among analogues such as microcystin-YR, microcystin-RR and nodularin. The computationally designed receptor was then used as a sensing element for the construction of sensor devices. A MIP-based piezoelectric sensor, capable of detecting 35 ~g rl of toxin in water, was developed. In order to improve the system sensitivity, the computational polymer was also used as a material in solid-phase extraction (SPE) for samples pre-concentration. The receptor was able to pre-concentrate up to 1,000 fold tap water samples spiked with only 1 J.1g rl of toxin. By combining MIP-based SPE and piezoelectric sensor an improved system with a minimum detectable concentration of toxin of 0.35 ~g rl was achieved. Encouraging preliminary results were also obtained in developing a MIP-based electrochemical sensor.