<p>Accurate and reliable ambiguity resolution (AR) is a primary approach to accelerate the convergence and improve the accuracy of precise point positioning (PPP). Currently, a very popular, maybe the most popular, method for addressing the PPP ambiguities is to decompose them into the wide lane and the “narrow lane” ambiguities. By introducing the satellite phase biases products, the wide line and the “narrow lane” ambiguities are gradually fixed in a cascade way. An alternative method of PPP-AR is by working directly with the raw ambiguities. Although the raw ambiguity method is very popular in scenario of real-time kinematic (RTK), especially in the case of short-baseline, it seems less popular in PPP practices. At present, it is not clear which method for PPP-AR is better, by working with the wide lane and the “narrow lane”, or with the raw ambiguities. Or perhaps they are indeed equivalent. In this contribution, a theoretical analysis on this subject is carried out. It is derived in a theorem that, based on the most widely used least-squares ambiguity decorrelation adjustment (LAMBDA), working with the raw ambiguities always fixes more ambiguities than working with the wide lane and the “narrow lane”. The theoretical result is validated by real GNSS data experiment. It further affirms that the raw ambiguity fixing has advantages in both convergence speed and positioning accuracy of PPP, thanks to more ambiguities are fixed. Therefore, it is more recommended to fix the raw ambiguity directly in PPP-AR, from the viewpoints of number of fixed ambiguities, convergence speed and positioning accuracy.</p>