To investigate the effect of magnesium (Mg) deficiency and excess on chlorophyll content and chlorophyll fluorescence characteristics of leaves, 2-year-old trees of Newhall navel orange (Citrus sinensis Osbeck) grafted on trifoliate orange (Poncirus trifoliata (L.) Raf.) rootstocks were transferred to pots containing a mixture of quartz sand-perlite (1:1, v/v) and irrigated with modified Hoagland-Aron nutrient solution containing three Mg concentrations: 0.1 mmol/L (Mg-deficiency), 1 mmol/L (control) or 10 mmol/L MgSO₄(Mg-excess). The relative chlorophyll content and chlorophyll fluorescence characteristics of leaves on different shoots were measured regularly in the three treatments. A portable chlorophyll meter (SPAD-502) was used to test the relative chlorophyll content of leaves. Simultaneously, the chlorophyll fluorescence characteristics of leaves were determined in situ by a pulse-amplitude modulated (PAM-2500) fluorometer. Results showed that Mg deficiency led to a more significant decline in chlorophyll content of 2-year-old autumn leaves than did Mg excess, whereas the latter caused a greater decline of 1-year-old leaves than did the former. In the 4^th month after Mg stress treatment, SPAD readings of leaves on 2-year-old autumn, 1-year-old spring, 1-year-old summer and 1-year-old late summer shoots in the Mg-deficient group decreased by 12.3% (P < 0.05), 10.5% (P < 0.05), 4.0% (P > 0.05) and 4.0% (P > 0.05), respectively, compared with the control. Those in the Mg-excess group were 107.3% (P < 0.05), 95.5% (P > 0.05), 88.9% (P < 0.05) and 92.5% (P < 0.05) of the control group, respectively. Moreover, both Mg-deficiency and excess reduced maximum quantum efficiency (F_v/F_m) and relative electron transport rate (rETR) of leaves on different shoots, with the impact of the former greater than that of the latter. In the 4^th month, compared with the control, F_v/F_m of 2-year-old autumn, 1-year-old spring, 1-year-old summer and 1year-old late summer leaves in the Mg-deficient group decreased by 13.9% (P < 0.05), 12.6% (P < 0.05), 2.9% (P > 0.05) and 2.0% (P > 0.05), respectively. Correspondingly, there was a decline of 0.5% (P > 0.05), 2.2% (P > 0.05), 3.4% (P > 0.05) and 1.5% (P > 0.05), respectively, in the Mg-excess group. Further investigation showed that the rETR_max of these leaves decreased by 35.7% (P < 0.05), 56.2% (P < 0.05), 32.6% (P < 0.05) and 15.2% (P < 0.05) in the Mg-deficient group, respectively, which corresponded to 110.1% (P < 0.05), 68.8% (P < 0.05), 87.2% (P < 0.05) and 84.5% (P < 0.05) of the control in the Mg-excess group. In addition, non-photochemical quenching (NPQ) of all leaves in the Mg-deficient group increased in the first 3 months and then declined dramatically in the 4^th month, although this was not as marked as in the Mg-excess group. In the 4^th month, the maximum NPQ of 2-year-old autumn, 1-year-old spring, 1-year-old summer and 1-year-old late summer leaves in the Mg-deficient group was lower than those in control, with a drop of 52.8% (P < 0.05), 26.6% (P < 0.05), 19.8% (P < 0.05) and 0.6% (P > 0.05), respectively. In the Mg-excess group, the peak value of NPQ of those leaves was 94.9% (P > 0.05), 110.3% (P > 0.05), 101.3% (P > 0.05) and 104.5% (P > 0.05) of the control group, respectively. Therefore, Mg-deficient plants exposed to high light intensity in summer would be subject to a reduction in photosynthetic capacity more easily than would Mg-excess plants. This indicated that Mg-deficient leaves would be susceptible to photoinhibition and photooxidation.