In this paper, different waveforms of frequency modulation (FM) chirp are investigated to analyze the accuracy of GNU radio-based frequency-modulated continuous wave (FMCW) radar for multiple target detection. The 3 waveforms used for the investigation as FM chirp are sinusoidal, triangular, and sawtooth waveforms. The analysis is performed by use of GNU radio referred as an open source software-define-radio project. There are 2 methods employed for the detection process; the first is real-condition simulation method and the second is USRP-based implementation method. In the analysis, some targets in different ranges are characterized using both methods to determine the accuracy of target range. By using FFT (Fast Fourier Transform) function from Matlab® to obtain the result in frequency domain, both methods show that the triangular waveform has the highest average accuracy, i.e. 95.73% for the 1 st method and 99.75% for the 2 nd method. The sawtooth waveform has the lower average accuracy than the triangular, i.e. 94.93 for the 1 st method and 98.33% for the 2 nd method, whilst the sinusoidal waveform has the lowest average accuracy, i.e. 92.60% for the 1 st method and 98.59% for the 2 nd method. From the result, it shows that the USRP-based implementation method has better average accuracy than the real-condition simulation method.