Landsat 8, the on going mission of the Landsat satellites that have provided over 40 years of images, continues to benefit long-term research. However, it is important to know if the spectral features of Landsat 8 are the same standard as previous Landsat imagery because Landsat 8 images have narrower bands, especially because of the normalized difference vegetation index (NDVI) calculation which is the most popular vegetation index. In this study, NDVI value is used to determine the vegetation index for calculating the spectral ratio between the value of NDVI Landsat 7 and the value of NDVI Landsat value of 8 , and it could be to calculate changes in land cover based on NDVI values .In this research the value of NDVI of Landsat 8 imagery compared with NDVI Landsat 7 derived from sample points 7 land cover classes are taken at the same month recording of Landsat 7 and Landsat 8 in Semarang. Seven (7) land cover classes used in this research are water, residential, industrial, barren land, garden, fields, and rice fields. For land cover changes in 2013, 2014 and 2015 in Semarang itself uses four classes of land cover: water, built-up land, infrequent vegetation and frequent vegetation.The results of spectral correlation between Landsat 7 and Landsat 8 with the method of NDVI in seven (7) classes of land cover in Semarang city by 2015 are obtained by the equation y = 0, 7171x + 0.0682 with R2 = 0,631 which has the definition that the influence of variable Landsat Landsat 7 against Landsat 8 is 63% and corellation coefficient (r) is 0,7943 which shows a strong positive correlation and, if Landsat 7 NDVI values increase, Landsat 8 NDVI values will increase too. On the results of Semarang land cover changes using NDVI showed a significant change from the year 2013-2015 on the extent of built up land class increase every year and vegetation is getting down every year.