The use of illegal fish drugs, such as malachite green (MG) in aquaculture has raised growing concerns over the safety of aquatic food products, and sensitive and rapid methods for detection of these drugs in fish muscle are much needed. Simplified extraction methods coupled with surface-enhanced Raman spectroscopy were used for analysis of MG and its metabolite leucomalachite green (LMG) in tilapia fillets containing 0–50 ng g−1 LMG or MG. Principal component analysis and partial least squares (PLS) regression were used for spectral data analyses. High sensitivity (MG or LMG could be detected at 1–2 ng g−1 levels) and PLS models with good predictability [root mean square error of validation (RMSEV) = 1.97–2.21 ng g−1, R2 = 0.984–0.988] were possible with suitable sample extraction methods. The simplest extraction method developed provided an RMSEV of 9.32 ng g−1 and R2 of 0.788, and although this was not ideal, detection of MG and LMG at 5 ng g−1 level with this method could reduce sample throughput time and provide for faster detection of LMG and MG. Copyright © 2012 John Wiley & Sons, Ltd.