The present chapter discusses the use of the total least squares (TLS) technique for parameter identification in benchmark functions and a real nonlinear photovoltaic (PV) model. The performance of TLS is then compared with that of the classic ordinary least squares (OLS) method using nine different metaheuristic optimisation algorithms, ABC, ALO, CA, DA, DE, GA, GWO, TLBO, and WOA. The study focuses on fitting four nonlinear functions, exponential decay, polynomial functions of degrees 3 and 5, and a sigmoid function, followed by the estimation of five PV cell parameters. The findings of the study demonstrate that TLS consistently yield more accurate parameter estimates and lower fitting errors than OLS across all methods. A statistical analysis encompassing 1,000 trials shows the impact of TLS on the variability of parameter estimates, as illustrated also for photovoltaic parameters estimation, especially when the two sensitive parameters Is and Rsh. Additionally, TLS achieved a lower mean absolute error (MAE) in modelling the PV cell’s current-voltage behaviour, with the best-performing combination being TLBO-TLS.