Microelectrode arrays (MEA) record extracellular local field potentials of cells adhered to the electrodes. A disadvantage is the limited signal-to-noise ratio. The state-of-the-art background noise level is about 10 muVpp. Furthermore, in MEAs low frequency events are filtered out. Here, we quantitatively analyze Au electrode/electrolyte interfaces with impedance spectroscopy and noise measurements. The equivalent circuit is the charge transfer resistance in parallel with a constant phase element that describes the double layer capacitance, in series with a spreading resistance. This equivalent circuit leads to a Maxwell-Wagner relaxation frequency, the value of which is determined as a function of electrode area and molarity of an aqueous KCl electrolyte solution. The electrochemical voltage and current noise is measured as a function of electrode area and frequency and follow unambiguously from the measured impedance. By using large area electrodes the noise floor can be as low as 0.3 muVpp. The resulting high sensitivity is demonstrated by the extracellular detection of C6 glioma cell populations. Their minute electrical activity can be clearly detected at a frequency below about 10 Hz, which shows that the methodology can be used to monitor slow cooperative biological signals in cell populations.