We detect 39 nm×10 nm gold nanorods using a microtoroid stabilized via the Pound-Drever-Hall method. Real-time detection is achieved with signal-to-noise ratios up to 12.2. These nanoparticles are a factor of three smaller in volume than any other nanoparticle detected using WGM sensing to date. We show through repeated experiments that the measurements are reliable, and verify the presence of single nanorods on the microtoroid surface using electron microscopy. At our current noise level, the plasmonic enhancement of these nanorods could enable detection of proteins with radii as small as a = 2 nm.Label-free single molecule detection has been an active area of research in optics during recent years, in part due to the emergence of whispering gallery mode (WGM) resonators as ultra-sensitive refractive index sensors 1,2 . Recently, however, several theoretical studies have emphasized that the predicted detection limit of current devices is well above that which is required for single molecule sensitivity 3 . For this reason many efforts have been made to improve the signal-to-noise ratio (SNR) and thereby reach the single molecule limit, including interferometry 4,5 , plasmonic enhancement 3,6,8,9 and frequency stabilization 10-12 . In this letter we build on these works, demonstrating real-time detection of gold (Au) nanorods with a silica microtoroid stabilized using the Pound-Drever-Hall (PDH) technique 13 . We detect 39 nm×10 nm nanorods with a SNR up to 12.2 and a resonator quality (Q) factor of 6×10 5 . These nanoparticles are ∼3 times smaller in volume than the smallest nanoparticles detected to date using the WGM sensing principle 4 .The essence of the PDH stabilization technique is in the measurement of an error signal which is fed back into the laser to supress fluctuations in frequency. The advantage of the technique is that it utilizes nulled lock-in detection, and the error signal is insensitive to a amplitude noise from the laser 13 . Because the laser is stabilized with respect to a reference cavity (in our case, a microtoroidal WGM resonator), the feedback loop ensures that the laser's frequency will follow any frequency shift δω in the cavity resonance, such as that experienced when a molecule binds to the microtoroid surface 1,3 . Fig. 1a shows a schematic illustrating the experimental setup. The setup consists of a 780 nm laser source (New Focus 6300-LN) coupled to a LiNbO 3 phase modulator (PM) and then to a silica microtoroid via a tapered optical fiber. A typical transmission spectrum of a microtoroid resonance in water is shown in Fig. 1b. The transmitted light is sent to a photodetector (D) and the a) Electronic