Pseudomonas fluorescens 9 and Bacillus subtilis 54, proposed as biofungicides to control Macrophomina phaseolina, a dangerous pathogen of soybean and other crops, were grown in vitro to evaluate their ability to produce metabolites with antifungal activity. The aim of the manuscript was to identify the natural compounds responsible for their antifungal activity. Only the culture filtrates of P. fluorescens 9 showed strong antifungal activity against M. phaseolina. Its organic extract contained phenazine and mesaconic acid (1 and 2), whose antifungal activity was tested against M. phaseolina, as well as Cercospora nicotianae and Colletotrichum truncatum, other pathogens of soybean; however, only compound 1 exhibited activity. The antifungal activity of compound 1 was compared to phenazine-1-carboxylic acid (PCA, 3), 2-hydroxyphenazine (2-OH P, 4), and various semisynthetic phenazine nitro derivatives in order to perform a structure–activity relationship (SAR) study. PCA and phenazine exhibited the same percentage of growth inhibition in M. phaseolina and C. truncatum, whereas PCA (3) showed lower activity against C. nicotianae than phenazine. 2-Hydroxyphenazine (4) showed no antifungal activity against M. phaseolina. The results of the SAR study showed that electron attractor (COOH and NO₂) or repulsor (OH) groups significantly affect the antifungal growth, as well as their α- or β-location on the phenazine ring. Both PCA and phenazine could be proposed as biopesticides to control the soybean pathogens M. phaseolina, C. nicotianae, and C. truncatum, and these results should prompt an investigation of their large-scale production and their suitable formulation for greenhouse and field applications.