Photoacoustic imaging can show the distribution of vessels and the degree of oxygen saturation, reliably supporting diagnoses of complaints such as cancer and articular rheumatism. Our handheld imaging system for percutaneous photoacoustic imaging and ultrasound imaging can take measurements easily. However, an important difficulty remains: reflection artifacts degrade the image quality. Reflection artifacts arise when photoacoustic waves from signal source reflect at the tissue boundary and are detected using a handheld probe. To address this difficulty, methods for identifying and eliminating reflection artifacts using single wavelength light have been developed.however, the detection accuracy of these methods is insufficient. On the other hand, tissue specificity of optical properties can achieve higher image quality when taking multi-wavelength measurements. We propose a method for identifying and eliminating reflection artifacts using the photoacoustic spectrum in multi-wavelength measurements. The photoacoustic signal intensity is wavelength-dependent and fluence-dependent. Reflection artifacts’ spectra are similar to that of the upper signal source. The spectrum of the signal source at the same depth as reflection artifact differs from the upper signal source. We took multi-wavelength measurements of a phantom mimicking vessels using piano wire in deaerated water including black ink or Intralipid of various consistencies. We then compared the photoacoustic spectra of piano wire and reflection artifacts by calculating root mean square (RMS). Results show that the reflection artifact spectrum and the upper signal source spectrum are more similar than the lower photoacoustic signal source spectrum. These analyses underscore the potential of this method for identifying and eliminating reflection artifacts.