Detrended fluctuation analysis (DFA) is a method used to quantify long-range power-law correlations in signals. The study investigates how nonstationarities—common in real data—can affect DFA results. Three types of nonstationarities are examined: (i) signals with segments removed due to data preprocessing, (ii) signals with random spikes, and (iii) signals with different local statistical properties. The study shows that introducing nonstationarities can lead to crossovers in scaling behavior, where the scaling exponent α changes with scale. The characteristics of these crossovers depend on factors such as the fraction and size of removed segments, spike concentration and amplitude, and the proportion of segments with different statistical properties. The study provides strategies for preprocessing data to minimize the effects of nonstationarities on scaling properties and interpreting DFA results for complex signals. It also demonstrates that the scaling behavior of nonstationary signals can be understood as a superposition of the scaling of individual components. The results show that for anti-correlated signals, removing segments can lead to a crossover from anti-correlated to uncorrelated behavior, while for correlated signals, the scaling remains unaffected. For signals with random spikes, the scaling behavior is a superposition of the signal and spike scaling. For signals with different local properties, the scaling behavior is a superposition of the scaling of individual segments. The study provides insights into how to analyze and interpret DFA results for complex signals with varying local characteristics.Detrended fluctuation analysis (DFA) is a method used to quantify long-range power-law correlations in signals. The study investigates how nonstationarities—common in real data—can affect DFA results. Three types of nonstationarities are examined: (i) signals with segments removed due to data preprocessing, (ii) signals with random spikes, and (iii) signals with different local statistical properties. The study shows that introducing nonstationarities can lead to crossovers in scaling behavior, where the scaling exponent α changes with scale. The characteristics of these crossovers depend on factors such as the fraction and size of removed segments, spike concentration and amplitude, and the proportion of segments with different statistical properties. The study provides strategies for preprocessing data to minimize the effects of nonstationarities on scaling properties and interpreting DFA results for complex signals. It also demonstrates that the scaling behavior of nonstationary signals can be understood as a superposition of the scaling of individual components. The results show that for anti-correlated signals, removing segments can lead to a crossover from anti-correlated to uncorrelated behavior, while for correlated signals, the scaling remains unaffected. For signals with random spikes, the scaling behavior is a superposition of the signal and spike scaling. For signals with different local properties, the scaling behavior is a superposition of the scaling of individual segments. The study provides insights into how to analyze and interpret DFA results for complex signals with varying local characteristics.