We report a new method for ultrasensitive detection of Cu(2+), which is based on changes in the tunnelling recognition current across self-assembled core-satellite gold nanoparticles (GNPs) networks functionalised with amino acids (l-cysteine). The addition of copper ions induces the formation of GNP/l-cysteine/Cu(2+)/l-cysteine/GNP molecular junctions and generates a significant decrease in the resistance through the networks. The networks are ultrasensitive to over ten orders range of copper ion concentrations.