Abstract As the first satellite altimetry mission operating in sar (delay-Doppler) mode, CryoSat-2 is an interesting mission to analyze its performance for water level monitoring over inland water bodies. It offers the opportunity to make comparison of sar and conventional altimeters by a multi-mode altimeter mounted on the same platform with a long repeat orbit. This comparison gives us more knowledge about the performance of the sar altimeter. Even tough it is not possible to perform it over same objects. In this paper we analyze the CryoSat-2 performance for water level monitoring via full- and sub-waveform retracking against in-situ gauge and L2 products of other satellite altimetry missions, e.g. Envisat and Jason-2. To this end, we retrack the full-waveforms and sub-waveforms with different empirical and physical retracking algorithms such as ocog, threshold, $\beta$-parameters and samosa3. We evaluate its capability in all measurement modes, i.e. lrm, sar and sarin, over inland water bodies located in different climatic zones. We selected study areas with different shapes and sizes. Relative to in situ measurements we find a precision of the CryoSat-2 lrm mode of $15\text{cm}$ rms over Qinghai lake (China) and $13\text{cm}$ over Erie lake (usa). This is an improvement over Envisat, yielding precision of $17\text{cm}$ in both cases. For the sar mode over Neagh lake (Northern Ireland) and Derg lake (Ireland) we obtain $15\text{cm}$ and $13\text{cm}$ where Envisat yields $28\text{cm}$ and $100\text{cm}$, respectively. The sarin mode’s precision is assessed over Nasser lake (Egypt) with $25\text{cm}$ rms and Athabasca lake (Canada) with $16\text{cm}$. Over these lakes Jason-2 achieved $54\text{cm}$ and Envisat $19\text{cm}$, respectively. The most precise results of CryoSat-2 are obtained with our retracking and sub-waveform selection scenarios. Comparing CryoSat-2 results from our retracking scenarios using L1b data with those results obtained from L2 products (data) of this mission shows an improvement of 4–17 cm. The minimum and maximum improvements belong to Erie and Derg lakes respectively, the largest and smallest lakes. From the waveform analysis over lakes with different shapes and sizes, we found that the first and the mean-all sub-waveforms (mean correction from all sub-waveforms) retracked with the threshold and samosa3 (only for sar mode) retrackers are appropriate to retrieve water level variation of small lakes and complex shaped lakes in this study. Over large lakes the full-waveform retracking leads to better results. In the case of icy-lake objects, sub-waveform retracking scenarios (the first and mean-all sub-waveforms) are more precise than the other scenarios. These are our findings over few samples, though more samples need to be analyzed to support them strongly.